Visual detection-oriented target detection model training method and target detection method

Abstract

The invention discloses a target detection model training method for visual detection and a target detection method. The training method comprises the following steps of: 1) selecting a plurality of annotated data and non-annotated data during each iterative training; 2) inputting the labeled image sample data into a target detection model for training, and obtaining a prediction result of each labeled image sample data; 3) loss calculation is carried out according to the prediction result and the corresponding label, and the loss Ls of supervision training is obtained; 4) performing weak enhancement and strong enhancement on each piece of unlabeled image sample data; 5) inputting the weakly enhanced sample data into the target detection model for prediction, and taking an obtained prediction result as a pseudo label corresponding to the strongly enhanced sample data; 6) inputting the strongly enhanced sample data into the target detection model for prediction, and performing loss calculation according to an obtained prediction result and a corresponding pseudo label to obtain a loss Lu of unsupervised training; and 7) adjusting parameters of the target detection model according to Ls and Lu.

Description

technical field [0001] The invention relates to the field of semi-supervised learning and target detection, in particular to a target detection model training method and target detection method oriented to visual detection. Background technique [0002] At present, most of the target detection models are fully supervised models, such as the target detection models of the "You Only LookOnce" (YOLO) series. The YOLO series is a single-stage target detection model that is biased towards speed. It unifies the target detection into a regression problem and greatly reduces the parameters required for prediction, so it can obtain the performance of high-speed detection, but the detection accuracy will be due to the reduction of the number of parameters. Decline. At present, the latest "You Only Look Once X, YOLOX version" of the YOLO series has the characteristics of fast detection speed and high precision, but it still requires a large amount of manually labeled data for training...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the problem that using a fully supervised target detection model in mobile phone appearance defect detection requires a lot of manpower and material resources for data labeling, the present invention proposes a visual detection-oriented target detection model training method and a target detection method; the present invention is based on semi-supervised The target detection training module Fix-YOLOX (Fixmatch-You Only Look OnceX), uses a small amount of labeled data for fully supervised training, and adds a semi-supervised training module using two methods of pseudo-labeling and consistency regularization, using unlabeled The data avoids overfitting of fully supervised training, improves the generalization of the model and the fault tolerance to labeled data

Images