A method and system for identifying the wearing of work clothes based on feature retrieval
A recognition method and technology for work clothes, applied in biometric recognition, character and pattern recognition, instruments, etc., can solve the problems of high time and computing cost, unable to recognize unknown objects, etc., to improve classification performance, save computing costs, save effect of time
Active Publication Date: 2021-01-15
ZHEJIANG LAB
View PDF8 Cites 1 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
However, the trained deep neural network can often only recognize known objects and cannot recognize unknown objects.
In the actual application scenario of work clothes recognition, if there are new work clothes or work clothes are replaced, it is necessary to re-collect the labeled data and retrain the network model, and the time and calculation costs are high.
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0033] Such as figure 1 As shown, the present invention provides a method for identifying the wearing of work clothes based on feature retrieval, comprising the following steps:
[0034] Step 1: Select pedestrian images from different angles of each type of work clothes and input them to the trained classification network to extract feature vectors from different angles of each type of work clothes to build a feature retrieval library;
[0035] Step 2: Input the image to be tested into the trained pedestrian detection network, crop the image according to the obtained pedestrian detection frame and input it to the trained classification network, obtain the feature vectors of all pedestrian detection frames in the image to be tested, and calculate each The feature vector of a pedestrian detection frame and the feature distance of the feature vector in the feature retrieval library;
[0036] Step 3: For each pedestrian detection frame, if the calculated feature distance from all...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a feature retrieval-based identification method and system for wearing work clothes. The method includes four steps of building a feature search database, detecting pedestrians, extracting work clothes features, and searching and identifying work clothes. Manually select representative image sets in different categories of work clothes, extract the features of work clothes through the classification network, and build a feature retrieval library; input the image to be tested into the pedestrian detection network to obtain the pedestrian area, and crop the pedestrian area in the image into the classification network extraction tool Clothing features; input the work clothes feature as the retrieval object into the feature retrieval library, calculate the distance between the object and each feature in the library, and finally output whether the object wears work clothes and the category of the work clothes worn. The invention can realize the flexible increase and decrease of the category of work clothes by simply rebuilding the feature retrieval library, effectively avoiding the huge cost of retraining the model, and can be used in the work clothes wearing recognition and alarm system in the factory workshop.
Description
technical field [0001] The invention relates to a feature retrieval-based recognition method for wearing work clothes and a related system, belonging to the technical field of industrial personnel safety monitoring. Background technique [0002] In industrial production and construction, safety has always been an eternal theme. The standard wearing of work clothes can provide certain protection for construction personnel when dangerous accidents occur, and effectively reduce the injuries suffered by construction personnel. However, construction workers do not wear work uniforms in a regular manner, which has caused great safety hazards. The traditional manual supervision method usually requires special staff to monitor the construction workers in factories, construction sites and other environments in real time through camera images. This method is costly and inefficient. In response to this problem, researchers have applied artificial intelligence technology to the field...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/00G06K9/46G06K9/62G06N3/04
CPCG06V40/10G06V10/40G06N3/045G06F18/23213G06F18/24G06F18/214
Inventor 章依依徐晓刚郑影王军黎晨阳朱岳江
Owner ZHEJIANG LAB