Restructuring of inference. Writing a class in C++ with two methods: init() to load configs and initialize the network and detect() to work with a sequence of frames (video)
To test the code, you can use my model.trt (disk.yandex.ru) (password: HFje), which I received after using my repository: Egorundel/yolov7_and_tensorrt-cpp-inference (github.com).
Note: this code does not save the result after detection, but displays the result during code execution