CN115063764A - Traffic signal lamp state estimation method and device and electronic equipment - Google Patents
Traffic signal lamp state estimation method and device and electronic equipment Download PDFInfo
- Publication number
- CN115063764A CN115063764A CN202210586997.8A CN202210586997A CN115063764A CN 115063764 A CN115063764 A CN 115063764A CN 202210586997 A CN202210586997 A CN 202210586997A CN 115063764 A CN115063764 A CN 115063764A
- Authority
- CN
- China
- Prior art keywords
- state
- information
- traffic signal
- signal lamp
- vehicle
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/584—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of vehicle lights or traffic lights
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention provides a method, a device and electronic equipment for estimating the state of a traffic signal lamp, which respond to the detection failure event of a target traffic signal lamp and firstly acquire the environmental information of the current vehicle and the historical detection information of the target traffic signal lamp; then, processing the environmental information and the historical detection information through a pre-trained state estimation model to obtain a preliminary estimation state of the target traffic signal lamp; and then determining the final estimation state of the target traffic signal lamp based on the historical detection information and the preliminary estimation state of the target traffic signal lamp. According to the method, under the condition that the traffic signal lamp is invalid in detection, the environment information and the historical detection information are processed through the state estimation model to obtain the initial estimation state of the traffic signal lamp, and the initial estimation state is further verified based on the historical monitoring information, so that the accuracy of state estimation of the traffic signal lamp is improved.
Description
Technical Field
The invention relates to the technical field of automatic driving, in particular to a method and a device for estimating the state of a traffic signal lamp and electronic equipment.
Background
Under the conditions that the lighting condition is poor, the picture quality is poor and the like, which are not beneficial to image detection and the like, the detection result of the traffic signal lamp obtained by detecting the vehicle-mounted picture shot by the camera is usually inaccurate, and the traffic accident is easily caused by adopting the detection result to guide the driving of the vehicle.
In the related art, a traffic signal lamp detection failure scene can be collected, and peripheral environment information can be analyzed to form a rule for determining traffic state information so as to guide the driving of the vehicle. However, the rule obtained in this way is generally complex and has poor consistency, and the obtained communication state information may not be accurate enough. The correct traffic light detection result can be used as a label, the surrounding environment information and the semantic map are used as input, and the neural network model is trained to be used for predicting the state of the traffic light, so that the traffic state information is determined. However, this approach requires a large amount of high quality training data, otherwise it is prone to misleading the neural network model, making the prediction results unreliable.
Disclosure of Invention
In view of this, the present invention provides a method, an apparatus and an electronic device for estimating a state of a traffic signal, so as to improve accuracy of a state estimation result of the traffic signal.
In a first aspect, an embodiment of the present invention provides a method for estimating a state of a traffic signal lamp, where the method includes: responding to a detection failure event of a target traffic signal lamp, and acquiring environmental information of a current vehicle and historical detection information of the target traffic signal lamp; the environment information comprises driving information of a current vehicle, attribute information and motion information of at least one moving object in a traffic scene of the current vehicle; the historical detection information comprises the passing state indicated by the target traffic signal lamp before the detection failure event of the target traffic signal lamp occurs; processing the environmental information and the historical detection information through a pre-trained state estimation model to obtain a preliminary estimation state of the target traffic signal lamp; and determining the final estimation state of the target traffic signal lamp based on the historical detection information and the preliminary estimation state of the target traffic signal lamp.
Optionally, in a first implementation manner of the first aspect of the present invention, the step of obtaining environmental information of the current vehicle includes: acquiring driving information of a current vehicle; the driving information comprises position information, speed information and direction information; acquiring attribute information and motion information of a moving object in a traffic scene of the current vehicle from a preset semantic map based on the position information of the current vehicle; the attribute information comprises the category information of the moving object; the motion information includes one or more of position information of the moving object, relative position information with the current vehicle, speed information, direction information, and relative position information with an intersection in the traffic scene.
Optionally, in a second implementation manner of the first aspect of the present invention, the state estimation model includes a first neural network and a second neural network; the first neural network is established based on a self-attention mechanism; the moving objects comprise pedestrians and vehicles; the attribute information includes a size of the vehicle; the motion information of the vehicle comprises the distance between the vehicle and the current vehicle, the lane where the vehicle is located and the position relation between the vehicle and the stop line; the step of processing the environment information and the historical detection information through a pre-trained state estimation model to obtain the preliminary estimation state of the target traffic signal lamp comprises the following steps: determining reference weight of the vehicle based on the size of the vehicle in the moving object and the motion information of the vehicle through a first neural network; and classifying the motion information of the pedestrian, the motion information of the vehicle, the reference weight of the vehicle and the historical detection information of the target traffic signal lamp through a second neural network to obtain the initial estimation state of the target traffic signal lamp.
Optionally, in a third implementation manner of the first aspect of the present invention, the state estimation model is obtained by training in the following manner: determining training data from a preset sample set; the training data comprises environmental information of the test vehicle, historical detection information of a traffic signal lamp corresponding to the environmental information and a current traffic state indicated by the traffic signal lamp; inputting environmental information of a test vehicle and historical detection information of a traffic signal lamp into an initial model to obtain a processing result output by the initial model; determining a loss value of the initial model based on the processing result and the current traffic state indicated by the traffic signal lamp; updating model parameters of the initial model based on the loss values; and continuing to execute the step of determining training data from preset sample data until the loss value is converged, and determining the initial model after the loss value is converged as a state estimation model.
Optionally, in a fourth implementation manner of the first aspect of the present invention, the sample set is determined by: acquiring multiple groups of running data of a preset test vehicle; the driving data comprises environmental information of the test vehicle, historical detection information of a traffic signal lamp corresponding to the environmental information and a current passing state indicated by the traffic signal lamp; the environmental information includes track information of the test vehicle; determining driving data meeting preset conditions in the plurality of groups of driving data as training data in the sample set; wherein the preset conditions include: and the track information of the test vehicle is matched with the current passing state of the traffic signal lamp corresponding to the environment information.
Optionally, in a fifth implementation manner of the first aspect of the present invention, the preliminary estimation state includes a traffic state indicated by a traffic light; the traffic state includes at least one of the following sub-states: a left-turn state, a straight-going state, a right-turn state and a turning state; the sub-state corresponds to pass or inhibit; the step of determining the final estimation state of the target traffic signal lamp based on the historical detection information and the preliminary estimation state of the target traffic signal lamp comprises the following steps: judging whether the process from the passing state indicated by the target traffic signal lamp to the passing state indicated by the traffic signal lamp in the preliminary estimation state accords with the traffic passing rule or not before the occurrence of the failure event is detected; if so, the preliminary estimated state is determined to be the final estimated state of the traffic signal lamp.
Optionally, in a sixth implementation manner of the first aspect of the present invention, the moving object includes at least one pedestrian or at least one vehicle; the attribute information includes category information of the moving object; the category information includes pedestrians or vehicles; the motion information comprises position information of the moving object; the position information comprises a sidewalk where a pedestrian is located or a lane where a vehicle is located; before inputting the environment information into the pre-trained traffic signal state estimation model, the method further comprises: dividing the moving object into a pedestrian group and a vehicle group based on the category information of the moving object; dividing the pedestrian group into at least one pedestrian lane group based on the sidewalk where the pedestrian is located in the pedestrian group; moving objects in the sidewalk group are positioned on the same sidewalk; dividing the vehicle group into at least one lane group based on the lanes where the vehicles are in the vehicle group; the moving objects in the lane group are positioned in the same lane; dividing attribute information and motion information of a moving object in the environment information into a plurality of data groups based on the sidewalk group and the lane group; one data set corresponds to a group of one-person lanes or a group of lanes.
Optionally, in a seventh implementation manner of the first aspect of the present invention, the motion information indicates a traffic state of the moving object; the preliminary estimated state comprises a traffic state indicated by a traffic light; the traffic state includes at least one of the following sub-states: a left-turn state, a straight-going state, a right-turn state and a turning state; the sub-state corresponds to pass or inhibit; before determining the final estimated state of the target traffic signal lamp based on the historical detection information and the preliminary estimated state of the target traffic signal lamp, the method further comprises: and for each sub-state, if the passing state of the moving object conflicts with the sub-state, removing the sub-state from the passing state indicated by the traffic signal lamp.
In a second aspect, an embodiment of the present invention provides a device for estimating a state of a traffic signal lamp, where the device includes: the environment information acquisition module is used for responding to a detection failure event of the target traffic signal lamp and acquiring the environment information of the current vehicle and the historical detection information of the target traffic signal lamp; the environment information comprises driving information of a current vehicle, attribute information and motion information of at least one moving object in a traffic scene of the current vehicle; the historical detection information comprises the passing state indicated by the target traffic signal lamp before the detection failure event of the target traffic signal lamp occurs; the state preliminary estimation module is used for processing the environment information and the historical detection information through a pre-trained state estimation model to obtain a preliminary estimation state of the target traffic signal lamp; and the state final estimation module is used for determining the final estimation state of the target traffic signal lamp based on the historical detection information and the preliminary estimation state of the target traffic signal lamp.
In a third aspect, an embodiment of the present invention provides an electronic device, which includes a processor and a memory, where the memory stores machine executable instructions that can be executed by the processor, and the processor executes the machine executable instructions to implement the above-mentioned method for estimating the state of a traffic signal lamp.
In a fourth aspect, embodiments of the present invention provide a machine-readable storage medium storing machine-executable instructions that, when invoked and executed by a processor, cause the processor to implement the above-described method for estimating a state of a traffic signal.
The embodiment of the invention has the following beneficial effects:
the embodiment of the invention provides a method, a device and electronic equipment for estimating the state of a traffic signal lamp, which respond to a detection failure event of a target traffic signal lamp and firstly acquire the environmental information of a current vehicle and the historical detection information of the target traffic signal lamp; then, processing the environmental information and the historical detection information through a pre-trained state estimation model to obtain a preliminary estimation state of the target traffic signal lamp; and then determining the final estimation state of the target traffic signal lamp based on the historical detection information and the preliminary estimation state of the target traffic signal lamp. According to the method, under the condition that the traffic signal lamp detection fails, the environment information and the historical detection information are processed through the state estimation model to obtain the preliminary estimation state of the traffic signal lamp, the preliminary estimation state is further verified based on the historical monitoring information, and the accuracy of the state estimation of the traffic signal lamp is improved.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.
In order to make the aforementioned and other objects, features and advantages of the present invention comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a flowchart of a method for estimating a state of a traffic signal according to an embodiment of the present invention;
FIG. 2 is a flow chart of another method for estimating a status of a traffic signal according to an embodiment of the present invention;
FIG. 3 is a flow chart of another method for estimating the status of a traffic signal according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a state estimation device of a traffic signal lamp according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
To make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Among the sensors used in autonomous vehicles, the camera is the only one with color perception capability. The task of detecting the status of traffic lights requires very accurate determination of the color and on/off status of each bulb, as well as accurate detection of its shape, orientation, count-down numbers, etc. to ensure that the current vehicle is legal, safe and comfortable. However, the camera as a device for collecting the state of the traffic light has many disadvantages, which are as follows:
1. the dynamic range is small, and overexposure is easy to happen when the light-emitting object is directly irradiated by sunlight or is excessively bright. It is also difficult to faithfully reflect the true brightness of a lamp with a relatively dark contrast.
2. Different cameras have large difference of imaging color tones and inaccurate reaction colors according to parameters such as exposure, shutter time and the like.
3. The field of view is narrow and multiple cameras are combined to provide sufficient coverage. But the two-phase machine interface is prone to truncation.
4. An LED (Light-Emitting Diode) traffic Light flashes periodically, and a stroboscopic problem occurs when the exposure frequency of a camera is low.
5. Moire effects can lead to color detection errors.
6. If the camera is placed at a low position, the camera is easily blocked by a cart, leaves, a telegraph pole, a billboard and the like, so that a complete traffic light cannot be seen.
7. Is easily affected by water drops, oil stains, dust and pollen, and causes imaging blurring.
In summary, the traffic light detection scheme based on vision has many defects which are difficult to overcome, so that the traffic light detection scheme cannot provide correct traffic light detection results in some scenes, and the current vehicle suddenly stops, runs a red light or is stuck.
However, when driving, a driver can generally determine how to drive according to the surrounding environment even if the driver cannot see the traffic light temporarily or cannot see the traffic light clearly. For example, as the vehicle flows away, no way to drive the vehicle behind is required; when vehicles corresponding to traffic lights in different directions pass at the intersection, the vehicles need to stop temporarily; the intersection without one person can pass through the lamp carefully when the lamp cannot be seen; the color of the traffic light of the road can be estimated approximately when the pedestrian light is seen.
In the related art, the schemes for estimating the state of the traffic light generally have the following two types:
collecting a traffic light detection failure scene, summarizing surrounding environment information, and forming a rule for guiding the behavior of the vehicle.
Collecting a large amount of intersection data with normal camera expression, using the traffic light detection result as a label, using the surrounding environment information and semantic map information as input features, and training a Recurrent Neural Network (RNN) for traffic state prediction.
The first scheme has the advantages of simplicity and directness, and has the disadvantages of complex and variable rules, possible contradiction, difficult maintenance and difficult coverage of long-tail scenes. The second scheme has the advantages of high precision and easy iteration maintenance, and has the defect that the model is easy to mislead when the training data quality is not high.
Based on this, the method, the device and the electronic device for estimating the state of the traffic signal lamp provided by the embodiment of the invention can be applied to the state estimation process of the traffic signal lamp in various scenes.
To facilitate understanding of the present embodiment, first, a detailed description is given of a method for estimating a state of a traffic signal disclosed in an embodiment of the present invention, as shown in fig. 1, the method includes the following steps:
102, responding to a detection failure event of a target traffic signal lamp, and acquiring environmental information of a current vehicle and historical detection information of the target traffic signal lamp; the environment information comprises driving information of a current vehicle, attribute information and motion information of at least one moving object in a traffic scene of the current vehicle; the historical detection information includes a traffic state indicated by the target traffic light before a detection failure event of the target traffic light occurs.
The detection failure event of the target traffic signal lamp can be that the quality of an image collected by a camera is poor, the collected image does not contain a signal lamp, or the state of the traffic signal lamp is detected to be wrong through the collected image, such as the events that all traffic signal lamps are not on at the same time.
After a detection failure event of the target traffic signal lamp is monitored, the environmental information of the current vehicle can be acquired. The environment information includes driving information of the current vehicle, such as driving speed, acceleration, lane where the vehicle is located, distance between the vehicle and a stop line, and attribute information and motion information of a moving object in the surrounding environment of the current vehicle. The attribute information can indicate that the moving object is a pedestrian, a non-motor vehicle or a motor vehicle, the size of the moving object and the like; the motion information is the lane or sidewalk where the moving object is located, the motion direction, the motion speed, the acceleration, whether the moving object passes a stop line or not and the like.
The driving information of the current vehicle can be generally obtained from a sensor system of the vehicle, such as the speed and the acceleration of the current vehicle can be read from a speed sensor, the distance between the lane where the current vehicle is located and the stop line can be identified from data collected by an image sensor or a laser radar arranged around the vehicle body, and the like.
Attribute information and motion information of a moving object can be generally obtained in combination with a sensor system and a semantic map of the vehicle. The sensor system of the vehicle can identify some surrounding moving objects and detect the speed, acceleration and other information of the moving objects. And the semantic map can display a moving object far away from the current vehicle, the information of the moving object in the set range of the current vehicle can be inquired in the semantic map, and the attribute information and the motion information of the moving object can be obtained comprehensively by combining the information of the moving object identified by the vehicle sensor system.
And 104, processing the environment information and the historical detection information through a pre-trained state estimation model to obtain a preliminary estimation state of the target traffic signal lamp.
The state estimation model is generally established based on a neural network or a deep learning theory, and may specifically be a cyclic neural network, a Transformer, a convolutional neural network, or the like. The state estimation model may classify the environmental information and the historical detection information, for example, when the speed of the vehicle in the same lane of the current vehicle is fast, the acceleration is positive acceleration, the relation with the stop line is that the stop line is not exceeded, the speed of the vehicle in the lane perpendicular to the lane of the current vehicle is slow or stopped, the acceleration is negative acceleration, the historical detection information of the traffic signal indicates that the lane normally passes, and the like, and then determine the preliminary estimation state of the target traffic signal based on the classification result. The preliminary estimated states of the target traffic signal lamps generally include traffic states of respective lanes, such as a left-turn lane, a straight-through lane, and a right-turn lane, such as normal traffic or no traffic.
The above example is an ideal situation, and in the actually acquired environment information, the passing states of different moving objects generally conflict with the passing state actually indicated by the traffic signal lamp. For example, a pedestrian or a vehicle may have a red light running behavior. Therefore, a large amount of environment information acquired under the condition that the recognition result of the traffic signal lamp is correct is required to be used as sample data, and the state estimation model is trained to obtain the state estimation model with high accuracy of state estimation of the traffic signal lamp.
Meanwhile, an attention mechanism can be set in the state estimation model, so that the weight of moving objects with different attributes in determining the preliminary estimation state of the target traffic signal lamp can be learned. Since pedestrians are less expensive to run red lights or have other behaviors that do not comply with traffic regulations, the pedestrian weight can often be reduced by learning in the training process of the model through an attention mechanism. The large vehicle is also easy to have some behaviors which do not comply with the traffic rules, and the weight of the large vehicle can be reduced by learning in the training process of the model through an attention mechanism.
And step 106, determining the final estimation state of the target traffic signal lamp based on the historical detection information and the preliminary estimation state of the target traffic signal lamp.
The historical detection information of the target traffic signal lamp is the passing state indicated by the target traffic signal lamp before the detection failure event of the target traffic signal lamp occurs. Generally, the traffic state indicated by the traffic light changes regularly, for example, a green light is usually turned on first when the traffic light goes straight, and then the green light is turned to the left when the traffic light goes straight to the red light. If the change rule formed by the detected preliminary estimation state of the target traffic signal lamp and the historical detection information does not accord with the preset rule, the preliminary estimation result can be considered to be invalid. For example, if the history detection information indicates that the left-turn lamp is green, the indication indicates that the left-turn traffic is allowed, and the preliminary estimation result indicates that the left-turn traffic is prohibited, and the straight traffic is normal, the change rule of the traffic signal lamp is not met, so that the preliminary estimation result of the traffic signal lamp is considered to be invalid. And if the change rule of the traffic signal lamp is met, determining the preliminary estimation state as the final estimation state.
The embodiment of the invention provides a state estimation method of a traffic signal lamp, which is used for responding to a detection failure event of a target traffic signal lamp, and firstly acquiring environmental information of a current vehicle and historical detection information of the target traffic signal lamp; then, processing the environmental information and the historical detection information through a pre-trained state estimation model to obtain a preliminary estimation state of the target traffic signal lamp; and then determining the final estimation state of the target traffic signal lamp based on the historical detection information and the preliminary estimation state of the target traffic signal lamp. According to the method, under the condition that the traffic signal lamp is invalid in detection, the environment information and the historical detection information are processed through the state estimation model to obtain the initial estimation state of the traffic signal lamp, and the initial estimation state is further verified based on the historical monitoring information, so that the accuracy of state estimation of the traffic signal lamp is improved.
The embodiment of the invention also provides another traffic signal lamp state estimation method which is realized on the basis of the method shown in the figure 1. The method mainly describes a specific process of acquiring environmental information of a current vehicle, a specific process of obtaining a preliminary estimation state of a target traffic signal lamp through a pre-trained state estimation model, and a specific process of training the state estimation model. As shown in fig. 2, the method comprises the steps of:
step S202, obtaining the driving information of the current vehicle; the driving information comprises position information, speed information and direction information. Specifically, it may be obtained from a sensing or monitoring system of the current vehicle.
Step S204, acquiring attribute information and motion information of a moving object in a traffic scene of the current vehicle from a preset semantic map based on the position information of the current vehicle; the attribute information comprises the category information of the moving object; the motion information includes one or more of position information of the moving object, relative position information with the current vehicle, speed information, direction information, and relative position information with an intersection in the traffic scene.
The moving object comprises at least one pedestrian or at least one vehicle; the attribute information is generally the category information of a moving object, such as a pedestrian or a vehicle; the motion information includes position information of a moving object and size information of a vehicle, such as a sidewalk where a pedestrian is located, or a lane where the vehicle is located, and the vehicle is a medium-sized vehicle, a large-sized vehicle, or a small-sized vehicle. After the environment information is obtained, the environment information can be classified, and the method can be specifically realized by the following steps:
(1) based on the category information of the moving object, the moving object is divided into a pedestrian group and a vehicle group.
(2) Dividing the pedestrian group into at least one pedestrian lane group based on the sidewalk where the pedestrian is located in the pedestrian group; the moving objects in the sidewalk group are located on the same sidewalk.
(3) Dividing the vehicle group into at least one lane group based on the lanes where the vehicles are in the vehicle group; the moving objects in the lane group are located in the same lane.
(4) Dividing attribute information and motion information of a moving object in the environment information into a plurality of data groups based on the sidewalk group and the lane group; one data set corresponds to a group of one-person lanes or a group of lanes.
The classification process can also be realized by a state estimation model, and the network structure of the state estimation model can be specifically designed according to requirements. Generally, a state estimation model may include a first neural network for determining weights and a second neural network for classification.
And step 206, determining the reference weight of the vehicle based on the size of the vehicle in the moving object and the motion information of the vehicle through the first neural network.
The first neural network can be established based on a self-attention mechanism or adopt a Transformer network structure. In general, a vehicle closer to the current vehicle is located in the same lane as the current vehicle, and the weight of the vehicle that does not pass through the stop line is higher.
And step 208, classifying the motion information of the pedestrian, the motion information of the vehicle, the reference weight of the vehicle and the historical detection information of the target traffic signal lamp through a second neural network to obtain the initial estimation state of the target traffic signal lamp.
The state estimation model can be trained in the following way:
(1) determining training data from a preset sample set; the training data comprises environmental information of the test vehicle, historical detection information of the traffic signal lamp corresponding to the environmental information and a current traffic state indicated by the traffic signal lamp.
The sample set may be determined by: firstly, acquiring a plurality of groups of running data of a preset test vehicle; the driving data comprises environmental information of the test vehicle, historical detection information of a traffic signal lamp corresponding to the environmental information and a current passing state indicated by the traffic signal lamp; the environmental information includes track information of the test vehicle; then determining the driving data meeting the preset conditions in the plurality of groups of driving data as training data in the sample set; wherein the preset conditions include: and the track information of the test vehicle is matched with the current passing state of the traffic signal lamp corresponding to the environment information.
(2) And inputting the environmental information of the test vehicle and the historical detection information of the traffic signal lamp into the initial model to obtain a processing result output by the initial model.
(3) Based on the processing result and the current traffic state indicated by the traffic signal lamp, a loss value of the initial model is determined.
(4) Updating model parameters of the initial model based on the loss values; and continuing to execute the step of determining training data from preset sample data until the loss value is converged, and determining the initial model after the loss value is converged as a state estimation model.
The preliminary estimation state of the target traffic light obtained by the state estimation model can comprise the passing state indicated by each traffic light; typically including at least one of the following sub-states: a left-turn state, a straight-going state, a right-turn state and a turning state; the sub-state corresponds to pass or inhibit.
The motion information of the moving object can indicate the passing state of the moving object; and for each sub-state in the preliminary estimation state, if the passing state of the moving object conflicts with the sub-state, removing the sub-state from the passing state indicated by the traffic signal lamp. And if the traffic state of the pedestrian on the sidewalk vertical to the lane of the current vehicle is normal traffic, and the lane where the current vehicle is located in the preliminary estimation state is normal traffic in the straight-going state, rejecting the straight-going state in the preliminary estimation state.
Step 210, judging whether the process from the traffic state indicated by the target traffic signal lamp to the traffic state indicated by the traffic signal lamp in the preliminary estimation state accords with traffic passing rules before the occurrence of the failure event; if yes, go to step S212; if not, the process is ended.
The traffic rule is usually that the straight traffic of lanes in the same direction is changed into left-turn traffic, then the traffic is forbidden, and whether the preliminary estimation state is reasonable or not needs to be judged according to the rule.
The preliminary estimated state is determined as the final estimated state of the traffic signal lamp, step 212.
The method improves the accuracy of state estimation of the traffic signal lamp and improves the traffic safety.
The embodiment of the invention also provides another traffic signal lamp state estimation method. The method is implemented on the basis of the method shown in fig. 1. The method aims to realize the following effects: the problem of generating and processing data and labels for training traffic flow estimation traffic lights in a large batch is solved; training a model for estimating the state of the traffic light by adopting non-manual labeled data, accurately predicting the walking/stopping and giving an unpredictable scene; and carrying out rule-based post-processing on the model output to make the output more in line with human intuition.
As shown in fig. 3, the method is specifically realized by the following steps:
step S302, screening data and determining training data.
Firstly, data near a road junction without taking over in road measurement is captured, wherein the fact that no taking over indicates that the detection result of the traffic light is effective. And then screening correct data (equivalent to the sample data set) for detecting the traffic lights by combining the relation between the traffic light state and the motion trail of the vehicle, namely whether the motion trail of the vehicle is reasonable in the traffic light state, and taking the detection result of the traffic lights as a label of training data.
Step S304, acquiring object tracking information, extracting the self characteristics of the vehicle from the object tracking information, and inquiring a semantic map to obtain the characteristic information of the object corresponding to the object tracking information.
The own characteristics of the vehicle (corresponding to the "driving information of the current vehicle") may include the direction, speed, acceleration, lane, etc. of the vehicle, and these own characteristics of the vehicle may be subsequently used as feature vectors to input a traffic light estimation model trained according to the screened data.
Specifically, after the semantic map is queried, objects such as vehicles and pedestrians can be grouped according to information such as lanes where the objects are located, target intersections, and traffic lights to which the objects belong, information such as speed, acceleration, object orientation, and size is extracted as feature information of the objects (equivalent to the above-mentioned "attribute information and motion information of moving objects"), and then the traffic light estimation model trained according to the screened data is input as a feature vector.
In step S306, the weight of the surrounding vehicle is determined by the self-attention mechanism.
The influence of the surrounding vehicles on the estimation of the traffic light state is different according to the distance of the vehicle, the lane where the vehicle is located, whether the stop line is passed or not, and the like. The peripheral vehicle features are subjected to a neural network self-attention mechanism, and weights can be learned so as to distinguish reference vehicles needing important attention.
In step S308, the traffic light state (corresponding to the "initial estimation state") is predicted by the recurrent neural network (corresponding to the "traffic light estimation model" described above).
And inputting all input characteristics and historical information thereof into a recurrent neural network for classification, and predicting the state of the traffic light.
And step S310, filtering the predicted traffic light state by adopting a preset rule.
And (4) verifying the output result of the neural network model by using a rule defined in advance, and removing obvious counterintuitive prediction. For example, a rule may be: the vehicle is prohibited from passing when pedestrians pass, is prohibited from passing when vehicles in the left direction and the right direction pass, is permitted to go straight when passing in the opposite direction, and the like.
In step S312, the traffic light detection result is fused to obtain the final traffic light state (corresponding to the "final estimated state" described above).
The abnormal state information (such as occlusion, reflection and camera failure) is accessed, and the traffic light state estimated by the traffic flow output in S6 is combined with the history of traffic light detection (especially the detection result before the abnormal condition occurs, which is equivalent to the above "history detection result").
The method provides a scheme for processing big data by a distributed cluster to quickly collect training data, designs a distributed data acquisition, screening and processing mode, and collects a large amount of data for training a traffic signal lamp estimation model; and environmental information extraction features are utilized, the features can comprise traffic light detection results and historical records thereof, camera failure information, historical tracks, orientations, speeds, accelerations and other information of surrounding vehicles and pedestrians, and the traffic light detection results and the historical tracks, the orientations, the speeds, the accelerations and the like are further classified by a recurrent neural network to obtain an estimated traffic light state. In the mode, the traffic light state is estimated by adopting the traffic flow and other environmental information, the problems that the automatic driving vehicle brakes suddenly at the traffic light intersection, runs the red light, is stuck and the like when the camera fails are solved, and the method can be used for estimating other intersection states where the traffic light cannot be seen clearly and helping to judge other vehicle intentions.
Corresponding to the above method embodiment, an embodiment of the present invention provides a device for estimating a state of a traffic signal, as shown in fig. 4, where the device includes:
an environment information obtaining module 402, configured to obtain environment information of a current vehicle and historical detection information of a target traffic light in response to a detection failure event of the target traffic light; the environment information comprises driving information of a current vehicle, attribute information and motion information of at least one moving object in a traffic scene of the current vehicle; the historical detection information comprises the passing state indicated by the target traffic signal lamp before the detection failure event of the target traffic signal lamp occurs;
a state preliminary estimation module 404, configured to process the environment information and the historical detection information through a pre-trained state estimation model to obtain a preliminary estimation state of the target traffic signal lamp;
and a state final estimation module 406, configured to determine a final estimated state of the target traffic light based on the historical detection information and the preliminary estimated state of the target traffic light.
Further, the environment information obtaining module is further configured to: acquiring driving information of a current vehicle; the driving information comprises position information, speed information and direction information; acquiring attribute information and motion information of a moving object in a traffic scene of the current vehicle from a preset semantic map based on the position information of the current vehicle; the attribute information comprises the category information of the moving object; the motion information includes one or more of position information of the moving object, relative position information with the current vehicle, speed information, direction information, and relative position information with an intersection in the traffic scene.
Further, the state estimation model includes a first neural network and a second neural network; the first neural network is established based on a self-attention mechanism; the moving objects comprise pedestrians and vehicles; the attribute information includes a size of the vehicle; the motion information of the vehicle comprises the distance between the vehicle and the current vehicle, the lane where the vehicle is located and the position relation between the vehicle and the stop line; the state preliminary estimation module is further to: determining reference weight of the vehicle based on the size of the vehicle in the moving object and the motion information of the vehicle through a first neural network; and classifying the motion information of the pedestrian, the motion information of the vehicle, the reference weight of the vehicle and the historical detection information of the target traffic signal lamp through a second neural network to obtain the initial estimation state of the target traffic signal lamp.
Further, the apparatus further comprises a model training module configured to: determining training data from a preset sample set; the training data comprises environmental information of the test vehicle, historical detection information of a traffic signal lamp corresponding to the environmental information and a current traffic state indicated by the traffic signal lamp; inputting environmental information of a test vehicle and historical detection information of a traffic signal lamp into an initial model to obtain a processing result output by the initial model; determining a loss value of the initial model based on the processing result and the current traffic state indicated by the traffic signal lamp; updating model parameters of the initial model based on the loss values; and continuing to execute the step of determining training data from preset sample data until the loss value is converged, and determining the initial model after the loss value is converged as a state estimation model.
Further, the apparatus further includes a sample set determining module configured to: acquiring multiple groups of running data of a preset test vehicle; the driving data comprises environmental information of the test vehicle, historical detection information of a traffic signal lamp corresponding to the environmental information and a current passing state indicated by the traffic signal lamp; the environmental information includes track information of the test vehicle; determining driving data meeting preset conditions in the plurality of groups of driving data as training data in the sample set; wherein the preset conditions include: and the track information of the test vehicle is matched with the current passing state of the traffic signal lamp corresponding to the environment information.
Further, the preliminary estimation state comprises a traffic state indicated by a traffic light; the traffic state includes at least one of the following sub-states: a left-turn state, a straight-going state, a right-turn state and a turning state; the sub-state corresponds to pass or inhibit; the state final estimation module is further configured to: judging whether the process from the passing state indicated by the target traffic signal lamp to the passing state indicated by the traffic signal lamp in the preliminary estimation state accords with the traffic passing rule or not before the occurrence of the failure event is detected; if so, the preliminary estimated state is determined to be the final estimated state of the traffic signal lamp.
Further, the moving object comprises at least one pedestrian or at least one vehicle; the attribute information includes category information of the moving object; the category information includes pedestrians or vehicles; the motion information comprises position information of the moving object; the position information comprises a sidewalk where a pedestrian is located or a lane where a vehicle is located; the above-mentioned device still includes: the first grouping module is used for dividing the moving object into a pedestrian group and a vehicle group based on the category information of the moving object; the pedestrian grouping module is used for dividing the pedestrian group into at least one pedestrian lane group based on the sidewalk where the pedestrian is located in the pedestrian group; moving objects in the sidewalk group are positioned on the same sidewalk; the vehicle grouping module is used for dividing the vehicle group into at least one lane group based on the lane where the vehicle is located in the vehicle group; the moving objects in the lane group are positioned in the same lane; the environment information grouping module is used for dividing attribute information and motion information of a moving object in the environment information into a plurality of data groups based on the sidewalk group and the lane group; one data set corresponds to a group of one-person lanes or a group of lanes.
Further, the motion information indicates the passing state of the moving object; the preliminary estimated state comprises a traffic state indicated by a traffic light; the traffic state includes at least one of the following sub-states: a left-turn state, a straight-going state, a right-turn state and a turning state; the sub-state corresponds to pass or inhibit; the device also comprises a state eliminating module used for: and for each sub-state, if the passing state of the moving object conflicts with the sub-state, removing the sub-state from the passing state indicated by the traffic signal lamp.
The embodiment also provides an electronic device, which comprises a processor and a memory, wherein the memory stores machine executable instructions capable of being executed by the processor, and the processor executes the machine executable instructions to realize the state estimation method of the traffic signal lamp.
Referring to fig. 5, the electronic device includes a processor 100 and a memory 101, the memory 101 stores machine executable instructions capable of being executed by the processor 100, and the processor 100 executes the machine executable instructions to implement the above-mentioned method for estimating the state of the traffic signal lamp.
Further, the electronic device shown in fig. 5 further includes a bus 102 and a communication interface 103, and the processor 100, the communication interface 103, and the memory 101 are connected through the bus 102.
The Memory 101 may include a high-speed Random Access Memory (RAM) and may also include a non-volatile Memory (non-volatile Memory), such as at least one disk Memory. The communication connection between the network element of the system and at least one other network element is realized through at least one communication interface 103 (which may be wired or wireless), and the internet, a wide area network, a local network, a metropolitan area network, and the like can be used. The bus 102 may be an ISA bus, PCI bus, EISA bus, or the like. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one double-headed arrow is shown in FIG. 5, but this does not indicate only one bus or one type of bus.
The present embodiments also provide a machine-readable storage medium storing machine-executable instructions that, when invoked and executed by a processor, cause the processor to implement the above-described traffic signal state estimation method.
The method, the device and the electronic device for estimating the state of the traffic signal lamp provided by the embodiment of the invention comprise a computer-readable storage medium storing program codes, wherein instructions included in the program codes can be used for executing the method described in the foregoing method embodiment, and specific implementation can refer to the method embodiment, which is not described herein again.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the system and the apparatus described above may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In addition, in the description of the embodiments of the present invention, unless otherwise explicitly specified or limited, the terms "mounted," "connected," and "connected" are to be construed broadly, e.g., as meaning either a fixed connection, a removable connection, or an integral connection; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meaning of the above terms in the present invention can be understood in specific cases for those skilled in the art.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present invention or portions thereof that contribute to the prior art may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the methods according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for convenience of description and simplicity of description, but do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art will understand that the following embodiments are merely illustrative of the present invention, and not restrictive, and the scope of the present invention is not limited thereto: those skilled in the art can still make modifications or changes to the embodiments described in the foregoing embodiments, or make equivalent substitutions for some features, within the scope of the disclosure; such modifications, changes or substitutions do not depart from the spirit and scope of the embodiments of the present invention, and they should be construed as being included therein. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (11)
1. A method of estimating a state of a traffic signal, the method comprising:
responding to a detection failure event of a target traffic signal lamp, and acquiring environmental information of a current vehicle and historical detection information of the target traffic signal lamp; the environment information comprises driving information of the current vehicle, attribute information and motion information of at least one moving object in a traffic scene of the current vehicle; the historical detection information comprises a traffic state indicated by a target traffic signal lamp before a detection failure event of the target traffic signal lamp occurs;
processing the environmental information and the historical detection information through a pre-trained state estimation model to obtain a preliminary estimation state of the target traffic signal lamp;
and determining the final estimation state of the target traffic signal lamp based on the historical detection information of the target traffic signal lamp and the preliminary estimation state.
2. The method according to claim 1, wherein the step of acquiring environmental information of the current vehicle includes:
acquiring driving information of a current vehicle; the driving information comprises position information, speed information and direction information;
acquiring attribute information and motion information of a moving object in a traffic scene of the current vehicle from a preset semantic map based on the position information of the current vehicle;
wherein the attribute information includes category information of the moving object; the motion information comprises one or more of position information of the moving object, relative position information with the current vehicle, speed information, direction information and relative position information with an intersection in the traffic scene.
3. The method of claim 1, wherein the state estimation model comprises a first neural network and a second neural network; the first neural network is established based on a self-attention mechanism; the moving object comprises a pedestrian and a vehicle; the attribute information includes a size of the vehicle; the motion information of the vehicle comprises the distance between the vehicle and the current vehicle, the lane where the vehicle is located and the position relation between the vehicle and the stop line;
the step of processing the environmental information and the historical detection information through a pre-trained state estimation model to obtain a preliminary estimation state of the target traffic signal lamp comprises the following steps:
determining, by a first neural network, a reference weight of a vehicle in the moving object based on a size of the vehicle, motion information of the vehicle;
and classifying the motion information of the pedestrian, the motion information of the vehicle, the reference weight of the vehicle and the historical detection information of the target traffic signal lamp through a second neural network to obtain the initial estimation state of the target traffic signal lamp.
4. The method of claim 1, wherein the state estimation model is trained by:
determining training data from a preset sample set; the training data comprise environmental information of a test vehicle, historical detection information of a traffic signal lamp corresponding to the environmental information and a current passing state indicated by the traffic signal lamp;
inputting the environmental information of the test vehicle and the historical detection information of the traffic signal lamp into an initial model to obtain a processing result output by the initial model;
determining a loss value of the initial model based on the processing result and the current traffic state indicated by the traffic signal lamp;
updating model parameters of the initial model based on the loss values; and continuing to execute the step of determining training data from preset sample data until the loss value is converged, and determining the initial model after the loss value is converged as a state estimation model.
5. The method of claim 4, wherein the sample set is determined by:
acquiring multiple groups of running data of a preset test vehicle; the driving data comprises environmental information of a test vehicle, historical detection information of a traffic signal lamp corresponding to the environmental information and a current passing state indicated by the traffic signal lamp; the environmental information includes trajectory information of the test vehicle;
determining driving data meeting preset conditions in the plurality of groups of driving data as training data in a sample set;
wherein the preset conditions include: and the track information of the test vehicle is matched with the current passing state of the traffic signal lamp corresponding to the environment information.
6. The method of claim 1, wherein the preliminary estimation state comprises a traffic state indicated by the traffic light; the traffic state includes at least one of the following sub-states: a left-turn state, a straight-going state, a right-turn state and a turning state; the sub-state corresponds to pass or inhibit;
the step of determining a final estimated state of the target traffic signal lamp based on the historical detection information of the target traffic signal lamp and the preliminary estimated state includes:
judging whether the process from the passing state indicated by the target traffic signal lamp to the passing state indicated by the traffic signal lamp in the preliminary estimation state accords with traffic passing rules or not before the occurrence of the failure event is detected;
and if so, determining the preliminary estimation state as the final estimation state of the traffic signal lamp.
7. The method of claim 1, wherein the moving object comprises at least one pedestrian or at least one vehicle; the attribute information includes category information of the moving object; the category information comprises pedestrians or vehicles; the motion information comprises position information of the moving object; the position information comprises a sidewalk where a pedestrian is located or a lane where a vehicle is located;
before inputting the environmental information into a pre-trained traffic signal state estimation model, the method further comprises:
dividing the moving object into a pedestrian group and a vehicle group based on the category information of the moving object;
dividing the pedestrian group into at least one pedestrian lane group based on the sidewalk where the pedestrian is located in the pedestrian group; moving objects in the sidewalk group are positioned on the same sidewalk;
dividing the vehicle group into at least one lane group based on the lane in which the vehicle is located in the vehicle group; the moving objects in the lane group are positioned in the same lane;
dividing attribute information and motion information of a moving object in the environment information into a plurality of data groups based on the sidewalk group and the lane group; one data set corresponds to one of the sidewalk group or the lane group.
8. The method of claim 1, wherein the motion information indicates a traffic status of the moving object; the preliminary estimated state comprises a traffic state indicated by the traffic light; the traffic state includes at least one of the following sub-states: a left-turn state, a straight-going state, a right-turn state and a turning state; the sub-state corresponds to pass or inhibit;
before determining a final estimated state of the target traffic signal based on the historical detection information of the target traffic signal and the preliminary estimated state, the method further comprises:
and for each sub-state, if the passing state of the moving object conflicts with the sub-state, removing the sub-state from the passing state indicated by the traffic signal lamp.
9. A state estimation apparatus of a traffic signal, characterized in that the apparatus comprises:
the system comprises an environmental information acquisition module, a traffic signal lamp detection failure detection module and a traffic signal lamp detection failure detection module, wherein the environmental information acquisition module is used for responding to a detection failure event of a target traffic signal lamp and acquiring environmental information of a current vehicle and historical detection information of the target traffic signal lamp; the environment information comprises driving information of the current vehicle, attribute information and motion information of at least one moving object in a traffic scene of the current vehicle; the historical detection information comprises a traffic state indicated by a target traffic signal lamp before a detection failure event of the target traffic signal lamp occurs;
the state preliminary estimation module is used for processing the environmental information and the historical detection information through a pre-trained state estimation model to obtain a preliminary estimation state of the target traffic signal lamp;
and the state final estimation module is used for determining the final estimation state of the target traffic signal lamp based on the historical detection information of the target traffic signal lamp and the preliminary estimation state.
10. An electronic device comprising a processor and a memory, the memory storing machine-executable instructions executable by the processor, the processor executing the machine-executable instructions to implement the method of traffic signal state estimation of any of claims 1-8.
11. A machine-readable storage medium having stored thereon machine-executable instructions that, when invoked and executed by a processor, cause the processor to implement the method of traffic signal state estimation of any of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210586997.8A CN115063764A (en) | 2022-05-26 | 2022-05-26 | Traffic signal lamp state estimation method and device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210586997.8A CN115063764A (en) | 2022-05-26 | 2022-05-26 | Traffic signal lamp state estimation method and device and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115063764A true CN115063764A (en) | 2022-09-16 |
Family
ID=83197697
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210586997.8A Pending CN115063764A (en) | 2022-05-26 | 2022-05-26 | Traffic signal lamp state estimation method and device and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115063764A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117475411A (en) * | 2023-12-27 | 2024-01-30 | 安徽蔚来智驾科技有限公司 | Signal lamp countdown identification method, computer readable storage medium and intelligent device |
-
2022
- 2022-05-26 CN CN202210586997.8A patent/CN115063764A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117475411A (en) * | 2023-12-27 | 2024-01-30 | 安徽蔚来智驾科技有限公司 | Signal lamp countdown identification method, computer readable storage medium and intelligent device |
| CN117475411B (en) * | 2023-12-27 | 2024-03-26 | 安徽蔚来智驾科技有限公司 | Signal lamp countdown identification method, computer readable storage medium and intelligent device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110619747A (en) | Intelligent monitoring method and system for highway road | |
| CN111723854B (en) | Expressway traffic jam detection method, equipment and readable storage medium | |
| CN111079680A (en) | Temporary traffic signal lamp detection method and device and automatic driving equipment | |
| WO2020007589A1 (en) | Training a deep convolutional neural network for individual routes | |
| KR20210097782A (en) | Indicator light detection method, apparatus, device and computer-readable recording medium | |
| CN114333344A (en) | Motor vehicle violation snapshot method and device and electronic equipment | |
| CN111649918A (en) | Method for monitoring turning on of high beam and continuous tracking of vehicle in real time | |
| CN111753612A (en) | Method and device for detecting sprinkled object and storage medium | |
| CN115049993B (en) | Vehicle abnormal stop monitoring method based on deep learning | |
| CN115063764A (en) | Traffic signal lamp state estimation method and device and electronic equipment | |
| CN113469075A (en) | Method, device and equipment for determining traffic flow index and storage medium | |
| JP7226368B2 (en) | Object state identification device | |
| CN112528944A (en) | Image identification method and device, electronic equipment and storage medium | |
| CN113313950A (en) | Method, device and equipment for detecting vehicle congestion and computer storage medium | |
| CN113468911A (en) | Vehicle-mounted red light running detection method and device, electronic equipment and storage medium | |
| CN112784817B (en) | Method, device and equipment for detecting lane where vehicle is located and storage medium | |
| CN116052119A (en) | Traffic light detection method and device, electronic equipment and computer readable storage medium | |
| CN113902999A (en) | Tracking method, device, equipment and medium | |
| CN113593253A (en) | Method and device for monitoring red light running of vehicle | |
| CN114241373A (en) | End-to-end vehicle behavior detection method, system, equipment and storage medium | |
| CN113085861A (en) | Control method and device for automatic driving vehicle and automatic driving vehicle | |
| CN112949470A (en) | Method, device and equipment for identifying lane-changing steering lamp of vehicle and storage medium | |
| CN113128264A (en) | Vehicle area determination method and device and electronic equipment | |
| CN110177222A (en) | A kind of the camera exposure parameter method of adjustment and device of the unused resource of combination vehicle device | |
| WO2023178510A1 (en) | Image processing method, device, and system and movable platform |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |