Open iscyy opened 2 years ago
Hi, the official standard models are the same with ultralytics. What I do is adding more submodules in this framework.
Hi, @Gumpest I have a question. Why does the standard yolov5l model choose a picture size of 608 during comparison, while the improved model uses a larger picture of 640. Is it because the accuracy of the improved model has dropped too much, and the data does not look intuitively As for the drop too much? Thank you author for the code
Hello author, if you run standard models such as yolov5m with your code, will the effect be exactly the same as running yolov5m with the official code? Or does your code have other optimizations besides the model structure?