I have a new repository to recommend. Roboflow Notebook contains all tutorials created by the team over the last few years. You'll find everything here - from fundamentals like ResNet32 to state-of-the-art Transformers. From classifications to image segmentation. And the best part is that the repository is constantly being developed, and new tutorials appear almost every week.
YOLOv5 is usually associated with object detection and is one of the most popular networks in the world for that task. Recently, image classification was added to YOLOv5, and it keeps getting better!
I hope you are doing well. I just released my new open-source library - onemetric. It is useful for evaluating computer vision projects for now, but I hope it will become the default benchmarking library for all ML-related projects. I would be very grateful if you could take a look. Potentially suggest some metrics from outside Computer Vision that might be useful.