Safety protector wearing detection method and system based on YOLOv5
A detection method and protective gear technology, applied in the YOLOv5-based safety protective gear wearing detection method and system field, can solve the problems of high accuracy, large safety hazards and accident risks, time-consuming and labor-intensive problems, and achieve fast speed and high accuracy , the effect of reducing the number of calculations
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0029] like figure 1 As shown, a kind of YOLOv5-based safety protective gear wearing detection method provided by the present invention includes:
[0030] Wherein, before step 101, it also includes:
[0031] Get the video data to be detected.
[0032] The video data to be detected is decomposed to obtain image data to be detected.
[0033] Step 101: Obtain image data to be detected.
[0034] Step 102: Input the image data to be detected into the YOLOv5 network for prediction to obtain a feature map; the feature map includes the target object category, the number of target objects, and the position of the detection frame. Wherein, step 102 specifically includes:
[0035] The Mosaic data enhancement algorithm is used to preprocess the image data to be detected to obtain preprocessed image data.
[0036] The preprocessed image data is input into the feature extraction network in the YOLOv5 network for feature extraction to obtain a salient feature map. Wherein, the describe...
Embodiment 2
[0057] The present invention also provides a more specific execution process of a YOLOv5-based safety protective gear wearing detection method, such as figure 2 and Figure 5 As shown, it specifically includes the following four steps:
[0058] 1. Generation of image datasets. Data images contain various objects such as construction workers, hard hats, seat belts, and safety clothing. Label various objects on each image, and divide the data set into training set, validation set and test set.
[0059] 2. Extraction and processing of image features. Build a feature extraction network model and a feature processing network model, perform multi-scale feature extraction and processing on the input image, and output the information for predicting the target object.
[0060] 3. Optimize the weight of the network model. Compare the information predicted by the network model with the labeled real data, calculate the loss function and perform backpropagation, and iteratively train...
Embodiment 3
[0154] The present invention also provides another embodiment of the YOLOv5-based safety protective gear wearing detection method in practical application, as follows:
[0155] 1: Generate an image dataset
[0156] The original image data set C is annotated by the image annotation tool LabelImg, and the six types of objects such as construction personnel, head, safety helmet, reflective clothing, ordinary clothing, and safety rope are placed in each image c i (1≤i≤n) is marked with a rectangular box. Then merge the original image and the labeled data to get the labeled data set L={l 1 , l 2 ,..., l n}, and merge the image data set C and the labeled data set L to obtain the data set D=(C, L), and finally divide the data set D into D train The training data set, D val Validation dataset, D test For the test data set, the ratio is set to 80:5:15.
[0157] to attach Image 6 For example, Image 6 (a) is the original image, Image 6 (b) is the labeled image. In the image, ...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More