CN114187664A - Rope skipping counting system based on artificial intelligence - Google Patents
Rope skipping counting system based on artificial intelligence Download PDFInfo
- Publication number
- CN114187664A CN114187664A CN202111527726.7A CN202111527726A CN114187664A CN 114187664 A CN114187664 A CN 114187664A CN 202111527726 A CN202111527726 A CN 202111527726A CN 114187664 A CN114187664 A CN 114187664A
- Authority
- CN
- China
- Prior art keywords
- rope
- skipping
- athlete
- module
- rope skipping
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Abstract
The invention discloses a rope skipping counting system based on artificial intelligence, which comprises: the industrial personal computer, the camera, the loudspeaker and the display screen are used for counting rope skipping athletes in the designated area; the industrial personal computer is provided with a data acquisition module, an instruction module, an athlete detection module, a preprocessing module, an illegal detection module, a counting module, an alarm module and a score output module. The invention can utilize artificial intelligence to realize intelligent counting of rope skipping counting, thereby reducing artificial counting errors and ensuring the accuracy and fairness of rope skipping counting.
Description
Technical Field
The invention belongs to the field of artificial intelligence, and particularly relates to a rope skipping counting system based on artificial intelligence.
Background
Skipping rope is a whole body aerobic exercise suitable for all ages, and at present, skipping rope is incorporated into the indispensable items of physical testing of primary and middle school students. Therefore, it becomes important to count the student skipping ropes.
At present, the skipping rope counting is generally carried out manually or by using skipping ropes with a counting function. When the number of rope skipping athletes is large, the precious time of a judge is delayed by adopting manual counting, and sometimes, when the rope skipping speed is high or the attention of the judge is not concentrated, counting errors easily occur. The counting rope skipping generally adopts a gear rotation mechanical counting principle, and can automatically record the rope skipping times. The method is simple and practical, can completely meet the counting requirement of daily movement, greatly reduces the inconvenience of manual counting, but due to the fact that the internal structure is not accurate enough, the counting mode sometimes has certain deviation.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, and provides a rope skipping counting system based on artificial intelligence, so that the intelligent counting of rope skipping counting can be realized by using the artificial intelligence, thereby reducing artificial counting errors and ensuring the accuracy and fairness of the rope skipping counting.
In order to achieve the purpose, the invention adopts the following technical scheme:
the invention relates to a rope skipping counting system based on artificial intelligence, which is characterized by comprising the following components: the system comprises an industrial personal computer, a camera, a loudspeaker and a display screen, and is used for counting rope skipping athletes in a designated area; the camera is arranged right ahead the designated area and used for shooting the skipping rope athlete in the forward direction, and the industrial personal computer is provided with: the system comprises a data acquisition module, an instruction module, an athlete detection module, a preprocessing module, an violation detection module, a counting module, an alarm module and a score output module;
after the instruction module sends a preparation signal through a loudspeaker and a display screen, the data acquisition module acquires an entry video of a jth rope skipping athlete in the designated area by using the camera and sends the entry video to the athlete detection module;
the athlete detection module detects the jth rope skipping athlete in the entry video, and when the jth rope skipping athlete is detected, the total rope skipping number q of the jth rope skipping athlete is initializedj0 and emptying the history stage state recording table Kj(ii) a The athlete detection module detects the distance between the key points of the shoulders of the jth skipping rope athlete as an initial distance d by using an OpenPose algorithm0jThe athlete detection module sends a detection completion signal to the instruction module;
the instruction module sends a take-off signal through a loudspeaker and a display screen after receiving the detection completion signal, and then starts timing;
the data acquisition module acquires a rope skipping video of the jth rope skipping athlete in the appointed area by using the camera according to the take-off signal and sends the rope skipping video to the preprocessing module;
the preprocessing module extracts a plurality of key frames in the rope skipping video and sends the key frames to the violation detection module;
the violation detection module detects the positions of the key points of the feet of the jth rope skipping player in each key frame to judge whether the violation detection module is in the designated standing area, if so, the detection is continued, otherwise, the line violation is judged and the line violation is reported through a loudspeaker;
the violation detection module judges whether a single-foot skipping rope exists according to the positions of the key points of the two feet of the jth skipping rope player, if so, the violation is judged and broadcasted through a loudspeaker, otherwise, the detection is continued;
the violation detection module judges whether a side skipping rope exists according to the distance between the key points of the shoulders of the jth rope skipping athlete, if so, the violation detection module judges that the side skipping rope exists and broadcasts the violation through a loudspeaker, and if not, the violation detection module sends a key frame passing the detection to the position detection module;
the position detection module detects the body position and the rope position of the jth rope skipping athlete in the key frame to judge the stage of rope skipping and stores the stage identifier into a historical stage state recording table KjPerforming the following steps;
the counting module records a table K according to the state of the historical stagejTotal number q of skipping ropes of jth skipping rope athletejCounting is carried out;
the instruction module sends a termination signal through a loudspeaker and a display screen after timing is finished;
the score output module acquires the total rope skipping number q of the counting module according to the termination signaljAnd broadcast through a loudspeaker and a display screen.
The rope skipping counting system based on artificial intelligence is also characterized in that the athlete detection module adopts a MaskRCNN target detection neural network to detect the jth rope skipping athlete; the MaskRCNN example segmentation neural network comprises the following steps: a depth residual error network ResNeXt-101, a feature pyramid network FPN, a region suggestion network RPN, an ROI Align layer and a classifier network;
the region proposal network RPN comprises: c. C1A convolution kernel size of k1×s1The step sizes of the convolution layers are s1，c2A convolution kernel size of k2×s2The step sizes of the convolution layers are s2；
The classifier network comprises three branches: mask branch of full convolution network, containing f1Bounding box regression branch of full connected layer, containing f2Softmax branches of all full connectivity layers;
inputting the incoming video into the Mask RCNN example segmentation neural network, and obtaining an image feature map through feature extraction of the depth residual error network ResNeXt-101 and the feature pyramid FPN;
inputting the image feature map into a regional suggestion network RPN, and respectively passing through c1A convolutional layer and c2Performing convolution processing on the convolution layers to correspondingly obtain anchor type and frame fine adjustment results;
the anchor type and the frame fine adjustment result pass through the ROIAlign layer together, a ROIAlign layer feature map is output, and the ROIAlign layer feature map is processed by utilizing a bilinear interpolation value to obtain a feature map after the region of interest is aligned;
inputting the feature map after aligning the interested regions into a classifier network, and obtaining a prediction Mask image through the full convolution processing of Mask branches;
the boundary frame regression branch performs boundary frame regression on the feature map and the prediction mask image after the region of interest is aligned to obtain a boundary frame coordinate;
the softmax branch performs softmax classification on the feature map after the region of interest is aligned to obtain a target category;
and taking the frame coordinates and the target category as a segmented target detection result to judge whether a rope skipping athlete is detected in the specified area.
The violation detection module detects the key frame by using an Open Pose algorithm to obtain 25 key point positions of the body of the rope skipping athlete and judges whether the rope skipping athlete is in a specified standing area or not according to the key point positions of both feet in the 25 key point positions; judging whether a single-foot skipping rope exists or not according to whether the distance between the vertical coordinates of the key point positions of the two feet exceeds a distance threshold value or not; according to the distance d between the 2 nd single-shoulder key point and the 8 th single-shoulder key pointjAt an initial distance d from0jAnd comparing to judge whether a side skipping rope exists.
The position detection module adopts a YOLOv5 target detection algorithm to extract the positions of a target frame and a rope of the body position in the key frame so as to judge the relative positions of the athlete and the rope; when the center of the rope is positioned at the upper half part of the target frame, the rope is in a rising stage, and the rising stage identifier is made to be '-1'; when the center position of the rope is in the middle part of the target frame, the rope is in the middle stage, and the middle stage identifier is made to be 0; when the center position of the rope is in the lower half of the target frame, it indicates that the rope is in the descent stage, and the descent stage flag is made to be "1".
The counting module counts the historical stage state recording table KjThe number of "-1", "0", "1" and "-1" appearing in succession in the phase identifier of (2) to obtain the total number of skipping ropes.
Compared with the prior art, the invention has the beneficial effects that:
1. the invention realizes the violation detection and intelligent counting of the rope skipping athletes by using the OpenPose algorithm and the Yolov5 target detection algorithm of the artificial intelligence technology, thereby enabling the counting result to be more accurate;
2. according to the invention, the human body posture key point data obtained by the OpenPose algorithm can effectively judge whether the rope skipping athlete has violation behaviors or not, and is not influenced by factors such as weather, illumination conditions, background and the like, so that the model has good robustness;
3. according to the method, the relative position of the rope and the athlete is judged through the YOLOv5 target detection algorithm, so that rope skipping counting is realized, the method is ensured to have high counting and identifying accuracy and precision, and the method has good application value.
Drawings
FIG. 1 is an OpenPose detection point bitmap of the present invention;
FIG. 2 is a detailed flow chart of the rope skipping counting of the present invention;
FIG. 3 is a diagram of a MaskRCNN example segmented neural network structure model according to the present invention.
Detailed Description
In this embodiment, referring to fig. 2, a rope skipping counting system based on artificial intelligence includes: the system comprises an industrial personal computer, a camera, a loudspeaker and a display screen, and is used for counting rope skipping athletes in a designated area; the camera has been arranged in the dead ahead in appointed area and is used for carrying out forward shooting to the rope skipping sportsman, is provided with on the industrial computer: the system comprises a data acquisition module, an instruction module, an athlete detection module, a preprocessing module, an violation detection module, a counting module, an alarm module and a score output module;
after the instruction module sends a preparation signal through a loudspeaker and a display screen, the data acquisition module acquires an entry video of a jth rope skipping athlete in the designated area by using a camera and sends the entry video to the athlete detection module;
the athlete detection module detects the jth rope skipping athlete in the entry video, and when the jth rope skipping athlete is detected, the total rope skipping number q of the jth rope skipping athlete is initializedj0 and emptying the history stage state recording table Kj(ii) a The athlete detection module uses the OpenPose algorithm to detect the key points of the shoulders of the jth rope skipping athlete, as shown in FIG. 1, namely, the coordinates (3 and 5 points) of the shoulders are obtained by using the Open Pose, so that the distance between the shoulders is calculated and used as the initial distance d0j. The athlete detection module sends a detection completion signal to the instruction module;
after receiving the detection completion signal, the instruction module sends a take-off signal through a loudspeaker and a display screen, and then starts timing;
the data acquisition module acquires a rope skipping video of the jth rope skipping athlete in the designated area by using the camera according to the take-off signal and sends the rope skipping video to the preprocessing module;
the method comprises the following steps that a preprocessing module extracts a plurality of key frames in a rope skipping video and sends the key frames to an illegal detection module;
the violation detection module detects the positions of the key points of the feet of the jth rope skipping player in each key frame to judge whether the violation detection module is in the designated standing area, if so, the detection is continued, otherwise, the line violation is judged and the line violation is reported through a loudspeaker;
the violation detection module judges whether a single-foot skipping rope exists according to the positions of the key points of the two feet of the jth skipping rope player, if so, the violation is judged and broadcasted through a loudspeaker, otherwise, the detection is continued;
the violation detection module judges whether a side skipping rope exists according to the distance between the key points of the shoulders of the jth rope skipping athlete, if so, the violation detection module judges that the side skipping rope exists and broadcasts the violation through a loudspeaker, and if not, the violation detection module sends a key frame passing the detection to the position detection module;
the position detection module detects the body position and the rope position of the jth rope skipping athlete in the key frame to judge the stage of rope skipping and stores the stage identifier into a historical stage state recording table KjPerforming the following steps;
the counting module records a table K according to the state of the historical stagejTotal number q of skipping ropes of jth skipping rope athletejCounting is carried out;
the instruction module sends a termination signal through a loudspeaker and a display screen after timing is finished;
the score output module acquires the total rope skipping number q of the counting module according to the termination signaljAnd broadcast through a loudspeaker and a display screen.
In specific implementation, referring to fig. 3, a MaskRCNN example segmentation neural network is constructed, and an athlete detection module in system operation detects the jth rope skipping athlete by adopting the MaskRCNN target detection neural network; MaskRCNN example segmentation neural network includes: a depth residual error network ResNeXt-101, a feature pyramid network FPN, a region suggestion network RPN, an ROI Align layer and a classifier network;
the region proposal network RPN comprises: c. C1A convolution kernel size of k1×s1The step sizes of the convolution layers are s1，c2A convolution kernel size of k2×s2The step sizes of the convolution layers are s2；
The classifier network includes three branches: mask branch of full convolution network, containing f1Bounding box regression branch of full connected layer, containing f2Softmax branches of all full connectivity layers;
inputting an incoming video into a MaskRCNN example segmentation neural network, and obtaining an image feature map through feature extraction of a depth residual error network ResNeXt-101 and a feature pyramid FPN;
inputting the image characteristic diagram into a regional suggestion network RPN, and respectively passing through c1A convolutional layer and c2Performing convolution processing on the convolution layers to correspondingly obtain anchor type and frame fine adjustment results;
in order to derive a smaller feature map from an ROI determined by an RPN (resilient packet network), aligning features acquired from each region of interest with the region of interest, passing the anchor type and frame fine adjustment results through a ROIAlign layer, outputting the ROIAlign layer feature map, and processing the ROIAlign layer feature map by using bilinear interpolation, wherein the bilinear interpolation is a better image scaling algorithm and fully utilizes four real pixel values around a virtual point in an original image to jointly determine one pixel value in a target image, namely the pixel value corresponding to the virtual position point can be estimated, and finally, the feature map after the region of interest is aligned is obtained;
inputting the feature map after aligning the interested areas into a classifier network, and expanding the aligned output dimension of the interested areas through the full convolution processing of Mask branches, so that the Mask can be predicted more accurately, and a predicted Mask image is obtained;
the boundary frame regression branch performs boundary frame regression on the feature map and the prediction mask image after the region of interest is aligned to obtain boundary frame coordinates;
performing softmax classification on the feature map after the region of interest is aligned by the softmax branch to obtain a target class;
and taking the frame coordinates and the target category as the segmented target detection result to judge whether the rope skipping athlete is detected in the specified area.
In this embodiment, the violation detection module detects the keyframe by using an Open Pose algorithm to obtain 25 key point positions of the body of the rope skipping athlete and judges whether the body is in the designated standing area according to the key point positions of both feet of the 25 key point positions; judging whether a single-foot skipping rope exists or not according to the fact whether the distance between the vertical coordinates of the positions of the key points of the two feet is different from a distance threshold value or not; according to the distance d between the 2 nd single-shoulder key point and the 8 th single-shoulder key pointjAt an initial distance d from0jAnd comparing to judge whether a side skipping rope exists.
In this embodiment, the position detection module extracts the positions of the target frame and the rope of the body position in the key frame by using the YOLOv5 target detection algorithm, and records the position coordinates (x _ P _ min, y _ P _ min, x _ P _ max, y _ P _ max) of the upper left corner and the lower right corner of the target frame, andrope center position coordinates (x _ R _ mid, y _ R _ mid) to determine the relative positions of the athlete and the rope; when the center of the rope is in the upper half of the target frame, i.e. when
Indicating that the rope is in a rising stage, and making a rising stage identifier be '-1'; when the center of the rope is positioned at the middle part of the target frame, i.e. when the rope is centered at the middle part
Indicating that the rope is in the middle stage, and making the identifier of the middle stage be '0'; when the cord is centered on the lower half of the target frame, i.e. when the cord is centered on the lower half
Indicating that the rope is in the descending stage, making the descending stage identifier be '1', and storing the stage identifier into the historical stage state recording table KjIn (1).
In this embodiment, the counting module counts the state record table K of the historical stagejIf the record table reaches the state of '1' from the state of '1' to the state of '1' through the state of '0', the counting standard is reached, and the number of skipping ropes q is countedjAnd adding one, emptying the record table, if the position does not reach the counting criterion, adding the position into the history record table, if the record table is full, emptying the record table, and restarting the recording, thereby obtaining the total number of the skipping ropes.
Claims (5)
1. The utility model provides a rope skipping counting system based on artificial intelligence which characterized in that includes: the system comprises an industrial personal computer, a camera, a loudspeaker and a display screen, and is used for counting rope skipping athletes in a designated area; the camera is arranged right ahead the designated area and used for shooting the skipping rope athlete in the forward direction, and the industrial personal computer is provided with: the system comprises a data acquisition module, an instruction module, an athlete detection module, a preprocessing module, an violation detection module, a counting module, an alarm module and a score output module;
after the instruction module sends a preparation signal through a loudspeaker and a display screen, the data acquisition module acquires an entry video of a jth rope skipping athlete in the designated area by using the camera and sends the entry video to the athlete detection module;
the athlete detection module detects the jth rope skipping athlete in the entry video, and when the jth rope skipping athlete is detected, the total rope skipping number q of the jth rope skipping athlete is initializedj0 and emptying the history stage state recording table Kj(ii) a The athlete detection module detects the distance between the key points of the shoulders of the jth skipping rope athlete as an initial distance d by using an OpenPose algorithm0jThe athlete detection module sends a detection completion signal to the instruction module;
the instruction module sends a take-off signal through a loudspeaker and a display screen after receiving the detection completion signal, and then starts timing;
the data acquisition module acquires a rope skipping video of the jth rope skipping athlete in the appointed area by using the camera according to the take-off signal and sends the rope skipping video to the preprocessing module;
the preprocessing module extracts a plurality of key frames in the rope skipping video and sends the key frames to the violation detection module;
the violation detection module detects the positions of the key points of the feet of the jth rope skipping player in each key frame to judge whether the violation detection module is in the designated standing area, if so, the detection is continued, otherwise, the line violation is judged and the line violation is reported through a loudspeaker;
the violation detection module judges whether a single-foot skipping rope exists according to the positions of the key points of the two feet of the jth skipping rope player, if so, the violation is judged and broadcasted through a loudspeaker, otherwise, the detection is continued;
the violation detection module judges whether a side skipping rope exists according to the distance between the key points of the shoulders of the jth rope skipping athlete, if so, the violation detection module judges that the side skipping rope exists and broadcasts the violation through a loudspeaker, and if not, the violation detection module sends a key frame passing the detection to the position detection module;
the position detection module is used for detecting the key frameThe body position and the rope position of the jth rope skipping athlete are detected to judge the stage of the rope skipping, and the stage identifier is stored in a historical stage state recording table KjPerforming the following steps;
the counting module records a table K according to the state of the historical stagejTotal number q of skipping ropes of jth skipping rope athletejCounting is carried out;
the instruction module sends a termination signal through a loudspeaker and a display screen after timing is finished;
the score output module acquires the total rope skipping number q of the counting module according to the termination signaljAnd broadcast through a loudspeaker and a display screen.
2. The artificial intelligence based rope skipping counting system of claim 1, wherein the athlete detection module uses a MaskRCNN target detection neural network to detect the jth rope skipping athlete; the MaskRCNN example segmentation neural network comprises the following steps: a deep residual error network ResNeXt-101, a feature pyramid network FPN, a region suggestion network RPN, a ROIAlign layer and a classifier network;
the region proposal network RPN comprises: c. C1A convolution kernel size of k1×s1The step sizes of the convolution layers are s1，c2A convolution kernel size of k2×s2The step sizes of the convolution layers are s2；
The classifier network comprises three branches: mask branch of full convolution network, containing f1Bounding box regression branch of full connected layer, containing f2Softmax branches of all full connectivity layers;
inputting the incoming video into the Mask RCNN example segmentation neural network, and obtaining an image feature map through feature extraction of the depth residual error network ResNeXt-101 and the feature pyramid FPN;
inputting the image feature map into a regional suggestion network RPN, and respectively passing through c1A convolutional layer and c2Performing convolution processing on the convolution layers to correspondingly obtain anchor type and frame fine adjustment results;
the anchor type and the frame fine adjustment result pass through the ROIAlign layer together, a ROIAlign layer feature map is output, and the ROIAlign layer feature map is processed by utilizing a bilinear interpolation value to obtain a feature map after the region of interest is aligned;
inputting the feature map after aligning the interested regions into a classifier network, and obtaining a prediction Mask image through the full convolution processing of Mask branches;
the boundary frame regression branch performs boundary frame regression on the feature map and the prediction mask image after the region of interest is aligned to obtain a boundary frame coordinate;
the softmax branch performs softmax classification on the feature map after the region of interest is aligned to obtain a target category;
and taking the frame coordinates and the target category as a segmented target detection result to judge whether a rope skipping athlete is detected in the specified area.
3. The rope skipping counting system based on artificial intelligence of claim 1, wherein the violation detection module detects the key frame by using an openpos algorithm to obtain 25 key point positions of the body of the rope skipping player and judges whether the rope skipping player is in the designated standing area according to the key point positions of both feet of the 25 key point positions; judging whether a single-foot skipping rope exists or not according to whether the distance between the vertical coordinates of the key point positions of the two feet exceeds a distance threshold value or not; according to the distance d between the 2 nd single-shoulder key point and the 8 th single-shoulder key pointjAt an initial distance d from0jAnd comparing to judge whether a side skipping rope exists.
4. The system according to claim 1, wherein the position detection module employs a YOLOv5 target detection algorithm to extract the target frame of body position in the key frame and the rope position to determine the relative position of the athlete and the rope; when the center of the rope is positioned at the upper half part of the target frame, the rope is in a rising stage, and the rising stage identifier is made to be '-1'; when the center position of the rope is in the middle part of the target frame, the rope is in the middle stage, and the middle stage identifier is made to be 0; when the center position of the rope is in the lower half of the target frame, it indicates that the rope is in the descent stage, and the descent stage flag is made to be "1".
5. The rope skipping counting system based on artificial intelligence as claimed in claim 4, wherein the counting module counts the historical stage state record table KjThe number of "-1", "0", "1" and "-1" appearing in succession in the phase identifier of (2) to obtain the total number of skipping ropes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111527726.7A CN114187664A (en) | 2021-12-14 | 2021-12-14 | Rope skipping counting system based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111527726.7A CN114187664A (en) | 2021-12-14 | 2021-12-14 | Rope skipping counting system based on artificial intelligence |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114187664A true CN114187664A (en) | 2022-03-15 |
Family
ID=80604975
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111527726.7A Pending CN114187664A (en) | 2021-12-14 | 2021-12-14 | Rope skipping counting system based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114187664A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115518330A (en) * | 2022-09-30 | 2022-12-27 | 安徽一视科技有限公司 | Rope skipping training system and method based on artificial intelligence |
| CN117079192A (en) * | 2023-10-12 | 2023-11-17 | 东莞先知大数据有限公司 | Method, device, equipment and medium for estimating number of rope skipping when personnel are shielded |
-
2021
- 2021-12-14 CN CN202111527726.7A patent/CN114187664A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115518330A (en) * | 2022-09-30 | 2022-12-27 | 安徽一视科技有限公司 | Rope skipping training system and method based on artificial intelligence |
| CN115518330B (en) * | 2022-09-30 | 2024-04-26 | 安徽一视科技有限公司 | Rope skipping training system and method based on artificial intelligence |
| CN117079192A (en) * | 2023-10-12 | 2023-11-17 | 东莞先知大数据有限公司 | Method, device, equipment and medium for estimating number of rope skipping when personnel are shielded |
| CN117079192B (en) * | 2023-10-12 | 2024-01-02 | 东莞先知大数据有限公司 | Method, device, equipment and medium for estimating number of rope skipping when personnel are shielded |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103871078B (en) | A kind of billiard ball batting key message detection method and system | |
| CN103413120B (en) | Tracking based on object globality and locality identification | |
| CN111414887B (en) | Secondary detection mask face recognition method based on YOLOV3 algorithm | |
| CN106682601B (en) | A kind of driver's violation call detection method based on multidimensional information Fusion Features | |
| CN104166841B (en) | The quick detection recognition methods of pedestrian or vehicle is specified in a kind of video surveillance network | |
| CN101477626B (en) | Method for detecting human head and shoulder in video of complicated scene | |
| CN102915438B (en) | The extracting method of a kind of video caption and device | |
| CN104778453B (en) | A kind of night pedestrian detection method based on infrared pedestrian's brightness statistics feature | |
| CN107316001A (en) | Small and intensive method for traffic sign detection in a kind of automatic Pilot scene | |
| CN102214309B (en) | Special human body recognition method based on head and shoulder model | |
| CN114187664A (en) | Rope skipping counting system based on artificial intelligence | |
| CN103020992B (en) | A kind of video image conspicuousness detection method based on motion color-associations | |
| CN107808376B (en) | Hand raising detection method based on deep learning | |
| CN102622584B (en) | Method for detecting mask faces in video monitor | |
| CN108399361A (en) | A kind of pedestrian detection method based on convolutional neural networks CNN and semantic segmentation | |
| CN108154102A (en) | A kind of traffic sign recognition method | |
| CN107491762A (en) | A kind of pedestrian detection method | |
| CN105913002B (en) | The accident detection method of online adaptive under video scene | |
| CN110210360A (en) | A kind of rope skipping method of counting based on video image target identification | |
| CN112270381B (en) | People flow detection method based on deep learning | |
| CN107944437B (en) | A kind of Face detection method based on neural network and integral image | |
| CN103065163B (en) | A kind of fast target based on static images detects recognition system and method | |
| CN101645137A (en) | Method for automatically detecting location of a football in long shot of football video | |
| CN113435336B (en) | Running intelligent timing system and method based on artificial intelligence | |
| CN109558790B (en) | Pedestrian target detection method, device and system |
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 |