CN106650641A - Traffic light positioning and identification method, device and system - Google Patents
Traffic light positioning and identification method, device and system Download PDFInfo
- Publication number
- CN106650641A CN106650641A CN201611104986.2A CN201611104986A CN106650641A CN 106650641 A CN106650641 A CN 106650641A CN 201611104986 A CN201611104986 A CN 201611104986A CN 106650641 A CN106650641 A CN 106650641A
- Authority
- CN
- China
- Prior art keywords
- traffic lights
- training sample
- traffic
- candidate recommendation
- image
- 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.)
- Granted
Links
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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
- G06F18/24137—Distances to cluster centroïds
- G06F18/2414—Smoothing the distance, e.g. radial basis function networks [RBFN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a traffic light positioning and identification method, device and system. The method includes the following steps that: an image to be detected and a convolutional neural network-based target detection and identification model are obtained; traffic light identification is performed on a traffic light identification region of the image to be detected through the convolutional neural network-based target detection and identification model; and traffic light information in the image to be detected is obtained, wherein the traffic light information contains the colors, coordinate positions and confidence degrees of traffic lights. According to the traffic light positioning and identifying method, device and system provided by the technical schemes of the invention adopted, the signals of the traffic lights are positioned and identified through the convolutional neural network-based target detection and identification model, and therefore, the traffic light positioning and identification method, device and system can be not only applicable for surrounding environment influence and but also be applicable to the positioning and identification of the signals of traffic lights of any shape, and thereby, improving the identification precision of the signals of the traffic lights and reducing a misjudgment rate.
Description
Technical field
The present invention relates to image identification technical field, more particularly to a kind of traffic lights positioning identifying method, device and
System.
Background technology
With the continuous improvement of economic level and living standards of the people, vehicle constantly increases in each big-and-middle small city.With
This simultaneously, thing followed traffic problems also increasingly receive much concern.It is automatic accurate during traffic violation is recognized
Identification traffic lights it is particularly important.Traditional traffic lights identifying schemes mainly use pixel change, and detection is special
Determine the color value changes in region to determine the state of traffic lights.For example user delimit fixed detection zone, if the area
Domain occurs red, then it is assumed that when red.But, the identifying schemes of this traffic lights are due to weather, the change in season, and thing
The characteristic that body expands with heat and contract with cold, the position of traffic lights can occur up and down, the movement of left and right, the position set when system initialization
It is not many times the position for finally detecting corresponding lamp to put.Simultaneously as the reason for illumination, imaging technique, it is actual red
Lamp is possibly shown as the color closer to amber light on imaging, causes the mistake for recognizing.This cross-color phenomenon is at night
Can become apparent from.
It is using the shape of traffic lights, such as circle to recognize that the technical scheme of traffic lights also has a kind of in prior art
The shapes such as shape, arrow carry out template matches, and the extracted region pixel value to matching in specific region, so as to recognize the face of lamp
Color.But, this technical scheme is different due to traffic lights, for example, justify taking the photograph for lamp, arrow lamp, and IMAQ
As the installation site of head and the installation site of traffic lights are had nothing in common with each other, cause the size of traffic lights between diverse location
Differ, it is difficult to find a common template that can cover various situations.And the program there is also the light occurred in technology one
According to cross-color phenomenon caused by, imaging technique, it is impossible to accurately carry out traffic lights identification.
Therefore, during inventor's exploitation traffic lights are recognized, find at least to exist in prior art and ask as follows
Topic:
Recognize that the technical scheme accuracy of identification of traffic lights is not high in prior art, False Rate is high, and poor universality, receives
External environment affects larger.
The content of the invention
In view of the above problems, it is proposed that the present invention so as to provide one kind overcome the problems referred to above or at least in part solve on
Problem is stated, the technical scheme is that what is be achieved in that:
On the one hand, the invention provides a kind of traffic lights positioning identifying method, including:
Obtain altimetric image to be checked and the target detection identification model based on convolutional neural networks;
By traffic signals of the target detection identification model based on convolutional neural networks to the altimetric image to be checked
Lamp detection zone carries out traffic lights identification;
Obtain traffic lights information in the altimetric image to be checked;The traffic lights information includes:Traffic lights
Color, coordinate position and confidence level.
Present invention also offers a kind of traffic lights positioning and recognizing device, including:
Information acquisition unit, the target detection identification model for obtaining altimetric image to be checked and based on convolutional neural networks;
Recognition unit, for by the target detection identification model based on convolutional neural networks to the mapping to be checked
The traffic lights detection zone of picture carries out traffic lights identification;
Information determination unit, for obtaining the altimetric image to be checked in traffic lights information;The traffic lights letter
Breath includes:The color of traffic lights, coordinate position and confidence level.
Present invention also offers a kind of traffic lights positioning identification system, including:As mentioned above traffic lights are positioned
Identifying device.
Technical scheme is by the target detection identification model based on convolutional neural networks to the mapping to be checked
As carrying out traffic lights identification, and obtain traffic lights information in the altimetric image to be checked.Due to described based on convolution god
The target detection identification model of Jing networks can learn to traffic lights coordinate position and traffic light color, so that
By the traffic lights in the altimetric image to be checked that the target detection identification model based on convolutional neural networks is identified
Information, that is, contain color, coordinate position and the confidence level of traffic lights.Therefore by the target based on convolutional neural networks
Detection identification model realizes the fixation and recognition of traffic lights, so as to solve prior art in traffic lights accuracy of identification not
Height, False Rate is high, and poor universality, and be affected by the external environment larger problem.
Description of the drawings
Fig. 1 is a kind of traffic lights positioning identifying method flow chart provided in an embodiment of the present invention;
Fig. 2 is another kind of traffic lights positioning identifying method flow chart provided in an embodiment of the present invention;
Fig. 3 is a kind of traffic lights positioning and recognizing device structural representation provided in an embodiment of the present invention;
Fig. 4 is a kind of traffic lights positioning identification system structural representation provided in an embodiment of the present invention.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with accompanying drawing to embodiment party of the present invention
Formula is described in further detail.
Algorithm based on convolutional neural networks.Wherein, fast area convolutional neural networks algorithm is current state-of-the-art use
One of deep learning neutral net in target detection.The fast area convolutional neural networks are made up of two parts network;One
Part is used to generate the region that target is likely to occur, and another part is used to carry out accurately target to be detected in the target area
Position and recognize.
If Fig. 1 is to show a kind of traffic lights positioning identifying method flow chart provided in an embodiment of the present invention;The party
Method includes:
101:Obtain altimetric image to be checked and the target detection identification model based on convolutional neural networks;
102:By traffic of the target detection identification model based on convolutional neural networks to the altimetric image to be checked
Signal lamp detection zone carries out traffic lights identification;
103:Obtain traffic lights information in the altimetric image to be checked;The traffic lights information includes:Traffic is believed
The color of signal lamp, coordinate position and confidence level.
Based on above example, as shown in Fig. 2 being another kind of traffic lights fixation and recognition provided in an embodiment of the present invention
Method flow diagram;The method includes:
201:Obtain the training sample with mark, treat training sample image and configuration parameter;The training sample of the mark
Coordinate position with one angle point of traffic signals lamp door, lamp door width high level and the color when headlight;
202:According to configuration parameter, the Candidate Recommendation region of training sample image is treated described in generation, and wait to instruct described in determining
Practice the coordinate position and its confidence level of traffic signals lamp door in the Candidate Recommendation region of sample image;The step implements process
It is as follows:The configuration parameter includes:Traffic signals lamp door length-width ratio and traffic lights collimation mark calibration information;The traffic signals
Lamp door standard information includes traffic signals lamp door coordinate position to be identified.
S21:According to the traffic signals lamp door length-width ratio, the Candidate Recommendation region of training sample image is treated described in generation;
S22:According to the traffic lights collimation mark calibration information, it is determined that the Candidate Recommendation region for treating training sample image
The coordinate position and its confidence level of middle traffic signals lamp door.The step is specially:
Technical solution of the present invention adopts the depth based on the convolutional neural networks quick detection model Faster R-CNN in region
Target detection identification model described in degree learning method off-line training based on convolutional neural networks.For example:When traffic signals lamp door
When standard information is that traffic lights are three perpendicular lamps in training sample, the ratio of width to height close 1 of the traffic lights:3, can be with
Size adjusting is carried out to image using this feature, it is ensured that the ratio of width to height is 1:3, and size is 20x60.Faster R-CNN depth
Learning model does not have strict requirements to the size and the ratio of width to height of target size, is able to detect that any width of reasonable range scale
Height compares target.According to the priori of traffic lights collimation mark calibration information, its ratio of width to height is set into 1:3,20x60 is sized to,
Model complexity can not only be reduced such that it is able to improve detection efficiency in the case where accuracy of detection is not sacrificed.The model is removed
The position of lamp door can be exported, outside the color of lamp, additionally it is possible to which (0-1,0 least to determine, 1 be to provide the confidence level of detection
Most determine).In real scene, as long as traffic lights lamp door just can be detected near 20x60, not being strict with size is
20x60。
203:According to the training sample with mark and traffic in the Candidate Recommendation region for treating training sample image
The coordinate position and its confidence level of signal lamp door, it is determined that the sample type in the Candidate Recommendation region for treating training sample image;
The sample type includes:Positive sample or negative sample;
204:According to traffic letter in training sample, the Candidate Recommendation region for treating training sample image with mark
The coordinate position and its confidence level of signal lamp frame and the sample type in the Candidate Recommendation region for treating training sample image, obtain institute
State based on the target detection identification model of convolutional neural networks.Wherein, the target detection based on convolutional neural networks is recognized
Model includes:Candidate Recommendation area detector and traffic lights classification and orientation device;It is as follows that the step implements process:
S41:According to traffic letter in training sample, the Candidate Recommendation region for treating training sample image with mark
The coordinate position and its confidence level of signal lamp frame and the sample type in the Candidate Recommendation region for treating training sample image, obtain institute
State Candidate Recommendation area detector;
S42:The Candidate Recommendation region of training sample image is treated according to Candidate Recommendation area detector output
The coordinate position and its confidence level of middle traffic signals lamp door and with mark training sample in mark with traffic signals lamp door
The coordinate position of one angle point, lamp door width high level and the color when headlight, obtain the traffic lights classification and orientation device.
It should be noted that before following steps S205, also including:Pre-set the traffic signals lamp inspection of altimetric image to be checked
Survey region;
205:By traffic of the target detection identification model based on convolutional neural networks to the altimetric image to be checked
Signal lamp detection zone carries out traffic lights identification;The step implements process:
S51:By the target detection identification model based on convolutional neural networks, the friendship of the altimetric image to be checked is determined
Candidate Recommendation region in ventilating signal lamp detection zone;
S52:Traffic lights identification is carried out in Candidate Recommendation region, described one angle point of traffic signals lamp door is obtained
Coordinate position, lamp door width high level, when the color and its confidence level of headlight;
S53:According to the coordinate position and confidence level of the traffic lights, the traffic lights detection zone is adjusted
Domain;Traffic lights detection zone after the adjustment will be recognized for the traffic lights of next round, so as to rear to be detected
The traffic lights identification of image;Concretely, it is exactly according to one angle point of traffic signals lamp door in the Candidate Recommendation region
Coordinate position and its confidence level dynamically adjust the size of its detection zone.For example:When the confidence in the Candidate Recommendation region
When spending very high (such as between 0.95-0.999), can be examined in the region of 40x 80 only around the present co-ordinate position
Survey.When confidence level be in higher level (such as between 0.85-0.95) when, can around present co-ordinate position 60x100 area
Detected in domain.When the confidence level is low (such as confidence level<0.85) target detection for being currently based on convolutional neural networks, is illustrated
Identification model does not know very much to testing result, needs to detect whole Candidate Recommendation region (100x 140 is bigger).Due to
Need to carry out convolution for Candidate Recommendation region based on the convolutional neural networks quick detection model Faster R-CNN in region, and
Time accounting is up to 90% or so, and image detection region is less, and speed is faster.So according to confidence level dynamic adjustment detection zone energy
It is enough effectively to improve detection efficiency, while also ensuring that the precision of detection.
It should be noted that using the model for training (i.e. through above step 201 to step 204 training based on convolution
The target detection identification model of neutral net) treat detection image positioned and the process that recognized before, the technology of the present invention side
Case can also pre-set the traffic lights detection zone of altimetric image to be checked;Due to the altimetric image to be checked of video acquisition device collection
Monitoring area it is larger, and the position of traffic lights has been relatively fixed, even if there is effect of expanding with heat and contract with cold, without departing from going out
Great scope (such as up and down within 2 times of traffic lights width).By described based on convolutional neural networks
Target detection identification model, before determining the Candidate Recommendation region of the altimetric image to be checked, the altimetric image to be checked can be with root
Determine the target detection based on convolutional neural networks according to the traffic lights detection zone for pre-seting altimetric image to be checked
The traffic lights detection zone (usually 5 times of lamp door width) of identification model.So, above step S51 passes through the base
In the target detection identification model of convolutional neural networks, the Candidate Recommendation region of the altimetric image to be checked is determined;It is determined here that
The Candidate Recommendation region of altimetric image to be checked is the traffic signals lamp inspection in the target detection identification model based on convolutional neural networks
Survey the Candidate Recommendation region determined in region.The area for needing the target detection identification model based on convolutional neural networks to recognize herein
Domain is substantially reduced, and can not only improve detection accuracy, moreover it is possible to shorten detection recognition time.
Meanwhile, adjust the size of traffic lights detection zone, it is ensured that the size of traffic lights lamp door is 20x60, so as to true
Protect model in the same size or close in application stage and training stage, and then improve system detectio accuracy.
206:Obtain traffic lights information in the altimetric image to be checked;The traffic lights information includes:Traffic is believed
The color of signal lamp, coordinate position and confidence level.It is as follows that the step implements flow process:
S61:Obtain the traffic lights information at least one Candidate Recommendation region of the altimetric image to be checked;
S62:The traffic lights information in confidence level highest Candidate Recommendation region is obtained, as the friendship of altimetric image to be checked
Ventilating signal lamp information.The concrete acquisition flow process of the step is as follows:At least one Candidate Recommendation region to the altimetric image to be checked
Interior traffic lights information is analyzed, and rejects possible flase drop.First, when appearance in the traffic lights identification process
There is weight between multiple Candidate Recommendation regions, and the lamp door of the traffic lights for the Candidate Recommendation region of two or more occur
It is folded, then need the overlapping area ratio for detecting lamp door for calculating two or more;Overlap between for example, two rectangle lamp doors, then
Described two lamp door overlapping area ratios are that the common factor of described two rectangle frames compares union;If described two lamp door overlapping area ratios
More than specific threshold (such as 0.5), then the relatively low rectangle lamp door of confidence level is eliminated, only retained highest confidence level correspondence rectangle
Lamp door result;If the confidence level of the lamp door is less than specific threshold (such as 0.5), the target based on convolutional neural networks
Detection identification model carries out the lamp door result of traffic lights identification to the altimetric image to be checked not know very much, flase drop occurs
Probability it is very big, now system will remove the corresponding Candidate Recommendation region of the lamp door, so as to ensure detection accuracy.
Secondly, carry out excluding abnormal situation according to the rule of traffic lights flicker, such as bright light order should be green
Lamp -->Amber light -->Red light -->Green light, because there may be completely black situation in the middle of the flicker of traffic lights, but will not go out
Show amber light situation bright immediately after when red, according to this rule some obvious flase drops can also be excluded.
Needs say that the above passes through green light -->Amber light -->Red light -->Green light phase exclude method should be
Screening to gathering image different images frame, i.e., complete erroneous frame according to the color change of the different lamps of information between frame and frame
Examination;And be then for Candidate Recommendation region different in same acquired image frames above with specific threshold investigation is compared
Investigation.
As shown in figure 3, being a kind of traffic lights positioning and recognizing device provided in an embodiment of the present invention;The device includes:
Information acquisition unit 301, the target detection for obtaining altimetric image to be checked and based on convolutional neural networks recognizes mould
Type;
Recognition unit 302, for by the target detection identification model based on convolutional neural networks to described to be checked
The traffic lights detection zone of altimetric image carries out traffic lights identification;
Information determination unit 303, for obtaining the altimetric image to be checked in traffic lights information;The traffic lights
Information includes:The color of traffic lights, coordinate position and confidence level.
It should be noted that the device also includes:
Parameter information acquiring unit, for obtaining the training sample with mark, treating training sample image and configuration parameter;Institute
State coordinate position of the training sample with one angle point of traffic signals lamp door, lamp door width high level and the color when headlight of mark;
Area generation unit, for treating the Candidate Recommendation region of training sample image described according to configuration parameter, generating, and
It is determined that in the Candidate Recommendation region for treating training sample image traffic signals lamp door coordinate position and its confidence level;
Type determining units, for being pushed away with the candidate for treating training sample image according to the training sample with mark
The coordinate position and its confidence level of traffic signals lamp door in region are recommended, it is determined that the Candidate Recommendation region for treating training sample image
Sample type;The sample type includes:Positive sample or negative sample;
Model acquiring unit, for being pushed away according to the training sample with mark, the candidate for treating training sample image
Recommend the coordinate position and its confidence level of traffic signals lamp door and the Candidate Recommendation region for treating training sample image in region
Sample type, obtains the target detection identification model based on convolutional neural networks;
Wherein, the configuration parameter includes:Traffic signals lamp door length-width ratio and traffic lights collimation mark calibration information;The area
Domain signal generating unit, specifically for treating that the candidate of training sample image pushes away described according to the traffic signals lamp door length-width ratio, generating
Recommend region;According to the traffic lights collimation mark calibration information, it is determined that handing in the Candidate Recommendation region for treating training sample image
The coordinate position and its confidence level of messenger lamp door;
Wherein, the target detection identification model based on convolutional neural networks includes:Candidate Recommendation area detector and
Traffic lights classification and orientation device;The model acquiring unit, specifically for according to the training sample with mark, described treat
The coordinate position and its confidence level of traffic signals lamp door and described training sample is treated in the Candidate Recommendation region of training sample image
The sample type in the Candidate Recommendation region of image, obtains the Candidate Recommendation area detector;According to the Candidate Recommendation region
The coordinate position and its confidence of traffic signals lamp door in the Candidate Recommendation region for treating training sample image of detector output
Degree and band mark training sample in mark the coordinate position with one angle point of traffic signals lamp door, lamp door width high level and ought
The color of headlight, obtains the traffic lights classification and orientation device.
Wherein, the device also includes:
Preset unit, for pre-seting the traffic lights detection zone of altimetric image to be checked;
The recognition unit is specifically for by the target detection identification model based on convolutional neural networks, determining institute
State Candidate Recommendation region in the traffic lights detection zone of altimetric image to be checked;Traffic lights are carried out in Candidate Recommendation region
Identification, obtains coordinate position, the lamp door width high level of one angle point of traffic signals lamp door, when the color and its confidence of headlight
Degree;According to the coordinate position and confidence level of the traffic lights, the traffic lights detection zone is adjusted;The adjustment
Traffic lights detection zone afterwards will be recognized for the traffic lights of next round, so as in the traffic of rear altimetric image to be checked letter
Signal lamp is recognized;
Described information determining unit, specifically at least one Candidate Recommendation region for obtaining the altimetric image to be checked
Traffic lights information;The traffic lights information in confidence level highest Candidate Recommendation region is obtained, as altimetric image to be checked
Traffic lights information.
As shown in figure 4, being a kind of traffic lights positioning identification system provided in an embodiment of the present invention;The system includes:
Traffic lights positioning and recognizing device as described above.
Technical scheme is by the target detection identification model based on convolutional neural networks to the mapping to be checked
As carrying out traffic lights identification, and obtain traffic lights information in the altimetric image to be checked.Due to described based on convolution god
The target detection identification model of Jing networks can learn to traffic lights coordinate position and traffic light color, so that
By the traffic lights in the altimetric image to be checked that the target detection identification model based on convolutional neural networks is identified
Information, that is, contain color, coordinate position and the confidence level of traffic lights.Therefore by the target based on convolutional neural networks
Detection identification model realizes the fixation and recognition of traffic lights, so as to solve prior art in traffic lights accuracy of identification not
Height, False Rate is high, and poor universality, and be affected by the external environment larger problem.
Presently preferred embodiments of the present invention is the foregoing is only, protection scope of the present invention is not intended to limit.It is all
Any modification, equivalent substitution and improvements made within the spirit and principles in the present invention etc., are all contained in protection scope of the present invention
It is interior.
Claims (10)
1. a kind of traffic lights positioning identifying method, it is characterised in that include:
Obtain altimetric image to be checked and the target detection identification model based on convolutional neural networks;
By traffic signals lamp inspection of the target detection identification model based on convolutional neural networks to the altimetric image to be checked
Surveying region carries out traffic lights identification;
Obtain traffic lights information in the altimetric image to be checked;The traffic lights information includes:The face of traffic lights
Color, coordinate position and confidence level.
2. method according to claim 1, it is characterised in that the method also includes:
Obtain the training sample with mark, treat training sample image and configuration parameter;The training sample of the mark carries traffic
The coordinate position of one angle point of signal lamp door, lamp door width high level and the color when headlight;
According to configuration parameter, the Candidate Recommendation region of training sample image is treated described in generation, and training sample figure is treated described in determining
The coordinate position and its confidence level of traffic signals lamp door in the Candidate Recommendation region of picture;
According to the training sample with mark and traffic signals lamp door in the Candidate Recommendation region for treating training sample image
Coordinate position and its confidence level, it is determined that the sample type in the Candidate Recommendation region for treating training sample image;The sample
Type includes:Positive sample or negative sample;
According to traffic signals lamp door in training sample, the Candidate Recommendation region for treating training sample image with mark
Coordinate position and its sample type in confidence level and the Candidate Recommendation region for treating training sample image, obtain described based on volume
The target detection identification model of product neutral net.
3. method according to claim 2, it is characterised in that the configuration parameter includes:Traffic signals lamp door length-width ratio
And traffic lights collimation mark calibration information;It is described according to configuration parameter, generate described in treat the Candidate Recommendation region of training sample image,
And treat the coordinate position and its confidence level step of traffic signals lamp door in the Candidate Recommendation region of training sample image described in determining,
Specifically include:
According to the traffic signals lamp door length-width ratio, the Candidate Recommendation region of training sample image is treated described in generation;
According to the traffic lights collimation mark calibration information, it is determined that traffic letter in the Candidate Recommendation region for treating training sample image
The coordinate position and its confidence level of signal lamp frame.
4. method according to claim 3, it is characterised in that the target detection based on convolutional neural networks recognizes mould
Type includes:Candidate Recommendation area detector and traffic lights classification and orientation device;It is described according to it is described with mark training sample,
The coordinate position of traffic signals lamp door and described training sample image is treated in the Candidate Recommendation region for treating training sample image
Candidate Recommendation region sample type, the step of obtain the target detection identification model based on convolutional neural networks, tool
Body includes:
According to traffic signals lamp door in training sample, the Candidate Recommendation region for treating training sample image with mark
Coordinate position and its sample type in confidence level and the Candidate Recommendation region for treating training sample image, obtain the candidate and push away
Recommend area detector;
Traffic letter in the Candidate Recommendation region of training sample image is treated according to Candidate Recommendation area detector output
The coordinate position and its confidence level of signal lamp frame and with mark training sample in mark with one angle point of traffic signals lamp door
Coordinate position, lamp door width high level and the color when headlight, obtain the traffic lights classification and orientation device.
5. the method according to Arbitrary Term in claim 1-4, it is characterised in that the method also includes:
Pre-set the traffic lights detection zone of altimetric image to be checked;
The traffic signals by the target detection identification model based on convolutional neural networks to the altimetric image to be checked
Lamp detection zone carries out traffic lights identification step to be included:
By the target detection identification model based on convolutional neural networks, the traffic lights of the altimetric image to be checked are determined
Candidate Recommendation region in detection zone;
Traffic lights identification is carried out in Candidate Recommendation region, the coordinate bit of one angle point of traffic signals lamp door is obtained
Put, lamp door width high level, when the color and its confidence level of headlight;
According to the coordinate position and confidence level of the traffic lights, the traffic lights detection zone is adjusted;The tune
Traffic lights detection zone after whole will be recognized for the traffic lights of next round, so as in the traffic of rear altimetric image to be checked
Signal lamp is recognized.
6. method according to claim 5, it is characterised in that traffic lights letter in the acquisition altimetric image to be checked
Breath;The traffic lights information includes:The step of color of traffic lights, coordinate position and confidence level, including:
Obtain the traffic lights information at least one Candidate Recommendation region of the altimetric image to be checked;
The traffic lights information in confidence level highest Candidate Recommendation region is obtained, as the traffic lights of altimetric image to be checked
Information.
7. a kind of traffic lights positioning and recognizing device, it is characterised in that include:
Information acquisition unit, the target detection identification model for obtaining altimetric image to be checked and based on convolutional neural networks;
Recognition unit, for by the target detection identification model based on convolutional neural networks to the altimetric image to be checked
Traffic lights detection zone carries out traffic lights identification;
Information determination unit, for obtaining the altimetric image to be checked in traffic lights information;The traffic lights packet
Include:The color of traffic lights, coordinate position and confidence level.
8. device according to claim 7, it is characterised in that the device also includes:
Parameter information acquiring unit, for obtaining the training sample with mark, treating training sample image and configuration parameter;The band
There are the coordinate position of one angle point of traffic signals lamp door, lamp door width high level and the color when headlight;
Area generation unit, for treating the Candidate Recommendation region of training sample image described according to configuration parameter, generating, and determines
The coordinate position and its confidence level of traffic signals lamp door in the Candidate Recommendation region for treating training sample image;
Type determining units, for according to the training sample with mark and the Candidate Recommendation area for treating training sample image
The coordinate position and its confidence level of traffic signals lamp door in domain, it is determined that the sample in the Candidate Recommendation region for treating training sample image
This type;The sample type includes:Positive sample or negative sample;
Model acquiring unit, for according to training sample, the Candidate Recommendation area for treating training sample image with mark
The sample in the coordinate position and its confidence level of traffic signals lamp door and the Candidate Recommendation region for treating training sample image in domain
Type, obtains the target detection identification model based on convolutional neural networks;
Wherein, the configuration parameter includes:Traffic signals lamp door length-width ratio and traffic lights collimation mark calibration information;The region life
Candidate Recommendation area into unit, specifically for treating training sample image described according to the traffic signals lamp door length-width ratio, generating
Domain;According to the traffic lights collimation mark calibration information, it is determined that traffic letter in the Candidate Recommendation region for treating training sample image
The coordinate position and its confidence level of signal lamp frame;
Wherein, the target detection identification model based on convolutional neural networks includes:Candidate Recommendation area detector and traffic
Signal lamp classification and orientation device;The model acquiring unit, specifically for according to the training sample with mark, described wait to train
The coordinate position and its confidence level of traffic signals lamp door and described training sample image is treated in the Candidate Recommendation region of sample image
Candidate Recommendation region sample type, obtain the Candidate Recommendation area detector;According to the Candidate Recommendation region detection
In the Candidate Recommendation region for treating training sample image of device output the coordinate position and its confidence level of traffic signals lamp door and
The coordinate position with one angle point of traffic signals lamp door that marks in training sample with mark, lamp door width high level and work as headlight
Color, obtain the traffic lights classification and orientation device.
9. according to the device described in claim 7 or 8, it is characterised in that the device also includes:
Preset unit, for pre-seting the traffic lights detection zone of altimetric image to be checked;
The recognition unit specifically for by the target detection identification model based on convolutional neural networks, it is determined that described treat
Candidate Recommendation region in the traffic lights detection zone of detection image;Traffic lights knowledge is carried out in Candidate Recommendation region
Not, coordinate position, the lamp door width high level of one angle point of traffic signals lamp door are obtained, when the color and its confidence level of headlight;
According to the coordinate position and confidence level of the traffic lights, the traffic lights detection zone is adjusted;After the adjustment
Traffic lights detection zone will be used for next round traffic lights recognize, so as in the traffic signals of rear altimetric image to be checked
Lamp is recognized;
Described information determining unit, specifically for the traffic at least one Candidate Recommendation region for obtaining the altimetric image to be checked
Signal information;The traffic lights information in confidence level highest Candidate Recommendation region is obtained, as the friendship of altimetric image to be checked
Ventilating signal lamp information.
10. a kind of traffic lights positioning identification system, it is characterised in that include:As claim 7 is appointed into claim 9
One traffic lights positioning and recognizing device of meaning.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611104986.2A CN106650641B (en) | 2016-12-05 | 2016-12-05 | A kind of traffic lights positioning identifying method, apparatus and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611104986.2A CN106650641B (en) | 2016-12-05 | 2016-12-05 | A kind of traffic lights positioning identifying method, apparatus and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106650641A true CN106650641A (en) | 2017-05-10 |
CN106650641B CN106650641B (en) | 2019-05-14 |
Family
ID=58819476
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611104986.2A Active CN106650641B (en) | 2016-12-05 | 2016-12-05 | A kind of traffic lights positioning identifying method, apparatus and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106650641B (en) |
Cited By (40)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107403169A (en) * | 2017-08-08 | 2017-11-28 | 上海识加电子科技有限公司 | Signal lamp detection recognition method and device |
CN107578439A (en) * | 2017-07-19 | 2018-01-12 | 阿里巴巴集团控股有限公司 | Generate the method, apparatus and equipment of target image |
CN107590456A (en) * | 2017-09-06 | 2018-01-16 | 张栖瀚 | Small micro- mesh object detection method in a kind of high-altitude video monitoring |
CN107689157A (en) * | 2017-08-30 | 2018-02-13 | 电子科技大学 | Traffic intersection based on deep learning can passing road planing method |
CN108132054A (en) * | 2017-12-20 | 2018-06-08 | 百度在线网络技术(北京)有限公司 | For generating the method and apparatus of information |
CN108189043A (en) * | 2018-01-10 | 2018-06-22 | 北京飞鸿云际科技有限公司 | A kind of method for inspecting and crusing robot system applied to high ferro computer room |
CN109389079A (en) * | 2018-09-30 | 2019-02-26 | 无锡职业技术学院 | A kind of traffic lights recognition methods |
CN109508635A (en) * | 2018-10-08 | 2019-03-22 | 哈尔滨理工大学 | A kind of traffic light recognition method based on TensorFlow combination multi-layer C NN network |
CN109508580A (en) * | 2017-09-15 | 2019-03-22 | 百度在线网络技术(北京)有限公司 | Traffic lights recognition methods and device |
CN109544955A (en) * | 2018-12-26 | 2019-03-29 | 广州小鹏汽车科技有限公司 | A kind of state acquiring method and system of traffic lights |
CN109657622A (en) * | 2018-12-21 | 2019-04-19 | 广东工业大学 | A kind of detection of traffic lights and recognition methods, device and equipment |
WO2019119356A1 (en) * | 2017-12-21 | 2019-06-27 | 华为技术有限公司 | Information detection method and mobile device |
CN110032916A (en) * | 2018-01-12 | 2019-07-19 | 北京京东尚科信息技术有限公司 | A kind of method and apparatus detecting target object |
CN110069986A (en) * | 2019-03-13 | 2019-07-30 | 北京联合大学 | A kind of traffic lights recognition methods and system based on mixed model |
CN110119725A (en) * | 2019-05-20 | 2019-08-13 | 百度在线网络技术(北京)有限公司 | For detecting the method and device of signal lamp |
CN110287244A (en) * | 2019-07-03 | 2019-09-27 | 武汉中海庭数据技术有限公司 | It is a kind of based on the traffic lights localization method repeatedly clustered |
CN110377670A (en) * | 2018-04-11 | 2019-10-25 | 腾讯大地通途(北京)科技有限公司 | A kind of method, apparatus, medium and the equipment of determining road element information |
CN110619307A (en) * | 2019-09-18 | 2019-12-27 | 广州文远知行科技有限公司 | Traffic light state determination method, device, equipment and storage medium |
WO2020001223A1 (en) * | 2018-06-29 | 2020-01-02 | 北京市商汤科技开发有限公司 | Method and device for traffic signal detection and intelligent driving, vehicle, and electronic device |
CN110826456A (en) * | 2019-10-31 | 2020-02-21 | 青岛海信网络科技股份有限公司 | Countdown board fault detection method and system |
CN110827552A (en) * | 2019-11-19 | 2020-02-21 | 北京小马慧行科技有限公司 | Vehicle running control method and device and vehicle |
CN110956153A (en) * | 2019-12-11 | 2020-04-03 | 安阳工学院 | Traffic signal lamp detection method and system for unmanned vehicle |
CN111079563A (en) * | 2019-11-27 | 2020-04-28 | 北京三快在线科技有限公司 | Traffic signal lamp identification method and device, electronic equipment and storage medium |
CN111127358A (en) * | 2019-12-19 | 2020-05-08 | 苏州科达科技股份有限公司 | Image processing method, device and storage medium |
CN111191595A (en) * | 2019-12-30 | 2020-05-22 | 上海眼控科技股份有限公司 | Vehicle identification detection method and device, computer equipment and readable storage medium |
CN111353511A (en) * | 2018-12-20 | 2020-06-30 | 富士通株式会社 | Number recognition device and method |
CN111428647A (en) * | 2020-03-25 | 2020-07-17 | 浙江浙大中控信息技术有限公司 | Traffic signal lamp fault detection method |
CN111505577A (en) * | 2020-04-27 | 2020-08-07 | 湖南大学 | Mobile vehicle positioning method based on visible light communication |
CN111738212A (en) * | 2020-07-20 | 2020-10-02 | 平安国际智慧城市科技股份有限公司 | Traffic signal lamp identification method, device, equipment and medium based on artificial intelligence |
CN111967368A (en) * | 2020-08-12 | 2020-11-20 | 广州小鹏车联网科技有限公司 | Traffic light identification method and device |
CN112288031A (en) * | 2020-11-18 | 2021-01-29 | 北京航空航天大学杭州创新研究院 | Traffic signal lamp detection method and device, electronic equipment and storage medium |
CN112289021A (en) * | 2020-09-24 | 2021-01-29 | 深圳一清创新科技有限公司 | Traffic signal lamp detection method and device and automatic driving automobile |
CN112908006A (en) * | 2021-04-12 | 2021-06-04 | 吉林大学 | Method for identifying state of road traffic signal lamp and counting down time of display |
CN112970030A (en) * | 2021-02-10 | 2021-06-15 | 华为技术有限公司 | Method and device for detecting traffic light |
CN113065466A (en) * | 2021-04-01 | 2021-07-02 | 安徽嘻哈网络技术有限公司 | Traffic light detection system for driving training based on deep learning |
WO2021134348A1 (en) * | 2019-12-30 | 2021-07-08 | 深圳元戎启行科技有限公司 | Traffic light state identification method and apparatus, computer device, and storage medium |
CN113536941A (en) * | 2021-06-18 | 2021-10-22 | 浙江吉利控股集团有限公司 | Training image generation, positioning detection network training and positioning detection method and device |
CN113723229A (en) * | 2021-08-16 | 2021-11-30 | 浙江大华技术股份有限公司 | Signal lamp detection method and device and computer readable storage medium |
CN114241438A (en) * | 2021-11-26 | 2022-03-25 | 电子科技大学 | Traffic signal lamp rapid and accurate identification method based on prior information |
CN114421613A (en) * | 2021-12-24 | 2022-04-29 | 国网湖北省电力有限公司宜昌供电公司 | Method for judging and processing running state of secondary protection equipment of transformer substation |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6760061B1 (en) * | 1997-04-14 | 2004-07-06 | Nestor Traffic Systems, Inc. | Traffic sensor |
CN103020603A (en) * | 2012-12-14 | 2013-04-03 | 上海交通大学 | Night signal light detection and identification method based on color space and selective voting mechanism |
CN103280114A (en) * | 2013-06-24 | 2013-09-04 | 电子科技大学 | Signal lamp intelligent control method based on BP-PSO fuzzy neural network |
CN103366190A (en) * | 2013-07-26 | 2013-10-23 | 中国科学院自动化研究所 | Method for identifying traffic sign |
CN105740751A (en) * | 2014-12-11 | 2016-07-06 | 深圳市赛为智能股份有限公司 | Object detection and identification method and system |
CN106023605A (en) * | 2016-07-15 | 2016-10-12 | 姹ゅ钩 | Traffic signal lamp control method based on deep convolution neural network |
-
2016
- 2016-12-05 CN CN201611104986.2A patent/CN106650641B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6760061B1 (en) * | 1997-04-14 | 2004-07-06 | Nestor Traffic Systems, Inc. | Traffic sensor |
CN103020603A (en) * | 2012-12-14 | 2013-04-03 | 上海交通大学 | Night signal light detection and identification method based on color space and selective voting mechanism |
CN103280114A (en) * | 2013-06-24 | 2013-09-04 | 电子科技大学 | Signal lamp intelligent control method based on BP-PSO fuzzy neural network |
CN103366190A (en) * | 2013-07-26 | 2013-10-23 | 中国科学院自动化研究所 | Method for identifying traffic sign |
CN105740751A (en) * | 2014-12-11 | 2016-07-06 | 深圳市赛为智能股份有限公司 | Object detection and identification method and system |
CN106023605A (en) * | 2016-07-15 | 2016-10-12 | 姹ゅ钩 | Traffic signal lamp control method based on deep convolution neural network |
Cited By (64)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107578439A (en) * | 2017-07-19 | 2018-01-12 | 阿里巴巴集团控股有限公司 | Generate the method, apparatus and equipment of target image |
CN107578439B (en) * | 2017-07-19 | 2020-04-28 | 创新先进技术有限公司 | Method, device and equipment for generating target image |
CN107403169A (en) * | 2017-08-08 | 2017-11-28 | 上海识加电子科技有限公司 | Signal lamp detection recognition method and device |
CN107403169B (en) * | 2017-08-08 | 2018-09-28 | 上海识加电子科技有限公司 | Signal lamp detection recognition method and device |
CN107689157A (en) * | 2017-08-30 | 2018-02-13 | 电子科技大学 | Traffic intersection based on deep learning can passing road planing method |
CN107689157B (en) * | 2017-08-30 | 2021-04-30 | 电子科技大学 | Traffic intersection passable road planning method based on deep learning |
CN107590456A (en) * | 2017-09-06 | 2018-01-16 | 张栖瀚 | Small micro- mesh object detection method in a kind of high-altitude video monitoring |
CN107590456B (en) * | 2017-09-06 | 2020-09-18 | 张栖瀚 | Method for detecting small and micro targets in high-altitude video monitoring |
CN109508580A (en) * | 2017-09-15 | 2019-03-22 | 百度在线网络技术(北京)有限公司 | Traffic lights recognition methods and device |
US11037005B2 (en) | 2017-09-15 | 2021-06-15 | Baidu Online Network Technology (Beijing) Co., Ltd. | Method and apparatus for identifying traffic light |
CN109508580B (en) * | 2017-09-15 | 2022-02-25 | 阿波罗智能技术(北京)有限公司 | Traffic signal lamp identification method and device |
CN108132054A (en) * | 2017-12-20 | 2018-06-08 | 百度在线网络技术(北京)有限公司 | For generating the method and apparatus of information |
WO2019119356A1 (en) * | 2017-12-21 | 2019-06-27 | 华为技术有限公司 | Information detection method and mobile device |
CN108189043A (en) * | 2018-01-10 | 2018-06-22 | 北京飞鸿云际科技有限公司 | A kind of method for inspecting and crusing robot system applied to high ferro computer room |
CN110032916A (en) * | 2018-01-12 | 2019-07-19 | 北京京东尚科信息技术有限公司 | A kind of method and apparatus detecting target object |
CN110377670A (en) * | 2018-04-11 | 2019-10-25 | 腾讯大地通途(北京)科技有限公司 | A kind of method, apparatus, medium and the equipment of determining road element information |
CN110377670B (en) * | 2018-04-11 | 2021-11-26 | 腾讯大地通途(北京)科技有限公司 | Method, device, medium and equipment for determining road element information |
KR102447352B1 (en) * | 2018-06-29 | 2022-09-23 | 베이징 센스타임 테크놀로지 디벨롭먼트 컴퍼니 리미티드 | Method and device for traffic light detection and intelligent driving, vehicle, and electronic device |
JP7111827B2 (en) | 2018-06-29 | 2022-08-02 | 北京市商▲湯▼科技▲開▼▲發▼有限公司 | Traffic light detection method, intelligent driving method and device, vehicle and electronic equipment |
JP2021519968A (en) * | 2018-06-29 | 2021-08-12 | 北京市商▲湯▼科技▲開▼▲發▼有限公司Beijing Sensetime Technology Development Co.,Ltd. | Traffic signal light detection methods, intelligent drive methods and devices, vehicles and electronic devices |
WO2020001223A1 (en) * | 2018-06-29 | 2020-01-02 | 北京市商汤科技开发有限公司 | Method and device for traffic signal detection and intelligent driving, vehicle, and electronic device |
CN110660254A (en) * | 2018-06-29 | 2020-01-07 | 北京市商汤科技开发有限公司 | Traffic signal lamp detection and intelligent driving method and device, vehicle and electronic equipment |
KR20200128145A (en) | 2018-06-29 | 2020-11-11 | 베이징 센스타임 테크놀로지 디벨롭먼트 컴퍼니 리미티드 | Methods and devices, vehicles, and electronic devices for traffic light detection and intelligent driving |
CN109389079B (en) * | 2018-09-30 | 2022-02-15 | 无锡职业技术学院 | Traffic signal lamp identification method |
CN109389079A (en) * | 2018-09-30 | 2019-02-26 | 无锡职业技术学院 | A kind of traffic lights recognition methods |
CN109508635B (en) * | 2018-10-08 | 2022-01-07 | 海南师范大学 | Traffic light identification method based on TensorFlow combined with multilayer CNN network |
CN109508635A (en) * | 2018-10-08 | 2019-03-22 | 哈尔滨理工大学 | A kind of traffic light recognition method based on TensorFlow combination multi-layer C NN network |
CN111353511B (en) * | 2018-12-20 | 2024-03-08 | 富士通株式会社 | Number recognition device and method |
CN111353511A (en) * | 2018-12-20 | 2020-06-30 | 富士通株式会社 | Number recognition device and method |
CN109657622A (en) * | 2018-12-21 | 2019-04-19 | 广东工业大学 | A kind of detection of traffic lights and recognition methods, device and equipment |
CN109544955A (en) * | 2018-12-26 | 2019-03-29 | 广州小鹏汽车科技有限公司 | A kind of state acquiring method and system of traffic lights |
CN110069986B (en) * | 2019-03-13 | 2021-11-02 | 北京联合大学 | Traffic signal lamp identification method and system based on hybrid model |
CN110069986A (en) * | 2019-03-13 | 2019-07-30 | 北京联合大学 | A kind of traffic lights recognition methods and system based on mixed model |
CN110119725A (en) * | 2019-05-20 | 2019-08-13 | 百度在线网络技术(北京)有限公司 | For detecting the method and device of signal lamp |
CN110119725B (en) * | 2019-05-20 | 2021-05-25 | 百度在线网络技术(北京)有限公司 | Method and device for detecting signal lamp |
CN110287244A (en) * | 2019-07-03 | 2019-09-27 | 武汉中海庭数据技术有限公司 | It is a kind of based on the traffic lights localization method repeatedly clustered |
CN110619307A (en) * | 2019-09-18 | 2019-12-27 | 广州文远知行科技有限公司 | Traffic light state determination method, device, equipment and storage medium |
CN110826456A (en) * | 2019-10-31 | 2020-02-21 | 青岛海信网络科技股份有限公司 | Countdown board fault detection method and system |
CN110827552A (en) * | 2019-11-19 | 2020-02-21 | 北京小马慧行科技有限公司 | Vehicle running control method and device and vehicle |
CN111079563A (en) * | 2019-11-27 | 2020-04-28 | 北京三快在线科技有限公司 | Traffic signal lamp identification method and device, electronic equipment and storage medium |
CN110956153A (en) * | 2019-12-11 | 2020-04-03 | 安阳工学院 | Traffic signal lamp detection method and system for unmanned vehicle |
CN110956153B (en) * | 2019-12-11 | 2023-02-07 | 安阳工学院 | Traffic signal lamp detection method and system for unmanned vehicle |
CN111127358A (en) * | 2019-12-19 | 2020-05-08 | 苏州科达科技股份有限公司 | Image processing method, device and storage medium |
CN111191595A (en) * | 2019-12-30 | 2020-05-22 | 上海眼控科技股份有限公司 | Vehicle identification detection method and device, computer equipment and readable storage medium |
WO2021134348A1 (en) * | 2019-12-30 | 2021-07-08 | 深圳元戎启行科技有限公司 | Traffic light state identification method and apparatus, computer device, and storage medium |
CN111428647B (en) * | 2020-03-25 | 2023-07-07 | 浙江中控信息产业股份有限公司 | Traffic signal lamp fault detection method |
CN111428647A (en) * | 2020-03-25 | 2020-07-17 | 浙江浙大中控信息技术有限公司 | Traffic signal lamp fault detection method |
CN111505577A (en) * | 2020-04-27 | 2020-08-07 | 湖南大学 | Mobile vehicle positioning method based on visible light communication |
CN111738212A (en) * | 2020-07-20 | 2020-10-02 | 平安国际智慧城市科技股份有限公司 | Traffic signal lamp identification method, device, equipment and medium based on artificial intelligence |
CN111738212B (en) * | 2020-07-20 | 2020-11-20 | 平安国际智慧城市科技股份有限公司 | Traffic signal lamp identification method, device, equipment and medium based on artificial intelligence |
CN111967368A (en) * | 2020-08-12 | 2020-11-20 | 广州小鹏车联网科技有限公司 | Traffic light identification method and device |
CN112289021A (en) * | 2020-09-24 | 2021-01-29 | 深圳一清创新科技有限公司 | Traffic signal lamp detection method and device and automatic driving automobile |
CN112288031A (en) * | 2020-11-18 | 2021-01-29 | 北京航空航天大学杭州创新研究院 | Traffic signal lamp detection method and device, electronic equipment and storage medium |
WO2022170540A1 (en) * | 2021-02-10 | 2022-08-18 | 华为技术有限公司 | Method and device for traffic light detection |
CN112970030A (en) * | 2021-02-10 | 2021-06-15 | 华为技术有限公司 | Method and device for detecting traffic light |
CN113065466A (en) * | 2021-04-01 | 2021-07-02 | 安徽嘻哈网络技术有限公司 | Traffic light detection system for driving training based on deep learning |
CN113065466B (en) * | 2021-04-01 | 2024-06-04 | 安徽嘻哈网络技术有限公司 | Deep learning-based traffic light detection system for driving training |
CN112908006B (en) * | 2021-04-12 | 2022-06-07 | 吉林大学 | Method for identifying state of road traffic signal lamp and counting down time of display |
CN112908006A (en) * | 2021-04-12 | 2021-06-04 | 吉林大学 | Method for identifying state of road traffic signal lamp and counting down time of display |
CN113536941A (en) * | 2021-06-18 | 2021-10-22 | 浙江吉利控股集团有限公司 | Training image generation, positioning detection network training and positioning detection method and device |
CN113723229A (en) * | 2021-08-16 | 2021-11-30 | 浙江大华技术股份有限公司 | Signal lamp detection method and device and computer readable storage medium |
CN114241438A (en) * | 2021-11-26 | 2022-03-25 | 电子科技大学 | Traffic signal lamp rapid and accurate identification method based on prior information |
CN114421613A (en) * | 2021-12-24 | 2022-04-29 | 国网湖北省电力有限公司宜昌供电公司 | Method for judging and processing running state of secondary protection equipment of transformer substation |
CN114421613B (en) * | 2021-12-24 | 2023-11-14 | 国网湖北省电力有限公司宜昌供电公司 | Method for judging and processing running state of secondary protection equipment of transformer substation |
Also Published As
Publication number | Publication date |
---|---|
CN106650641B (en) | 2019-05-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106650641B (en) | A kind of traffic lights positioning identifying method, apparatus and system | |
CN106407981B (en) | A kind of licence plate recognition method, apparatus and system | |
CN108596081B (en) | Vehicle and pedestrian detection method based on integration of radar and camera | |
CN108898047B (en) | Pedestrian detection method and system based on blocking and shielding perception | |
CN107330453B (en) | Pornographic image identification method based on step-by-step identification and fusion key part detection | |
CN107705288B (en) | Infrared video detection method for dangerous gas leakage under strong interference of pseudo-target motion | |
CN108764257A (en) | A kind of pointer instrument recognition methods of various visual angles | |
CN105868691A (en) | Urban vehicle tracking method based on rapid region convolutional neural network | |
CN108537775A (en) | A kind of cancer cell tracking based on deep learning detection | |
CN104268538A (en) | Online visual inspection method for dot matrix sprayed code characters of beverage cans | |
CN105512640A (en) | Method for acquiring people flow on the basis of video sequence | |
CN106778737A (en) | A kind of car plate antidote, device and a kind of video acquisition device | |
CN108052887A (en) | A kind of doubtful illegal land automatic recognition system and method for merging SLAM/GNSS information | |
Almagambetov et al. | Mobile standards-based traffic light detection in assistive devices for individuals with color-vision deficiency | |
CN110910341A (en) | Transmission line corrosion area defect detection method and device | |
CN107016353B (en) | A kind of integrated method and system of variable resolution target detection and identification | |
CN115841488B (en) | PCB hole inspection method based on computer vision | |
CN102129685B (en) | Method for detecting irregular circle based on Gauss pyramid decomposition | |
CN116148801B (en) | Millimeter wave radar-based target detection method and system | |
CN115661615A (en) | Training method and device of image recognition model and electronic equipment | |
CN109255279A (en) | A kind of method and system of road traffic sign detection identification | |
CN104613869A (en) | Method and system for detecting elliptical hole group based on maximum inscribed circle | |
CN103324957A (en) | Identification method and identification device of state of signal lamps | |
CN114022820A (en) | Intelligent beacon light quality detection method based on machine vision | |
CN108563997A (en) | It is a kind of establish Face datection model, recognition of face method and apparatus |
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 | ||
CB02 | Change of applicant information |
Address after: 8 floors of Block E, No.2 Building, 9 Yuan, Fenghao East Road, Haidian District, Beijing 100094 Applicant after: Wen'an Beijing intelligent technology Limited by Share Ltd Address before: 100085 4th Floor, No. 1 Courtyard, Shangdi East Road, Haidian District, Beijing Applicant before: Wen'an Beijing intelligent technology Limited by Share Ltd |
|
CB02 | Change of applicant information | ||
GR01 | Patent grant | ||
GR01 | Patent grant |