Target detection can be said to be the most important technology in the application of deep learning in the industry today, not one of them. I have read a lot of articles about the engineering practice of target detection, and I feel that I have gained a lot. The following contains some links to articles that have been carefully read, which can be used as a reference for target detection practice.
Tricks in deep neural network model training (principle and code summary)
A collection of tricks in the target detection competition
[Target detection practice] How many images does the detector need at least?
The direction that can be improved in the target detection practice - know almost
[Hydrology 3] Some engineering methods to improve model speed/accuracy-Knowledge
New reflections in model training - Zhihu
Improve the idea of small target detection
Feature fusion skills in inventory target detection (summarized according to YOLO v4)
Solve small target detection! Summary of Multiscale Methods
Incredible target detection trick: train a few more epochs, and get a better model on average