pytorch-yolov3 | pytorch implementation of YOLOv3 for real-time webcam | Computer Vision library

I packaged (using Pyinstaller) a small variant of the Minimalistic Yolo github repo, found Here, the packaging was done using pyinstaller to run the object detection as a server using Flask.

So while attempting to run the server, it only works when running from Anaconda Prompt (Which is where i wrote the pyinstaller command) other than that, the following error occur.

Error i Get while running from (exe,Cmd,PowerShell) is:

I'm using an existing PyTorch-YOLOv3 architecture and training it to recognize a custom dataset through google colab for a research manuscript. Basically I want to use the object detection algorithm to count the number of objects for two classes in an image.

I've been told that for my purpose, I should generate validation/training curves for the model and create a confusion matrix to evaluate the classifier element of the trained model. I have an idea to modify the training script to output training metrics to a csv file during the training, but I'm not familiar with how to create a confusion matrix to evaluate the trained model.

Additionally, in the field of computer vision, what kind of metrics/figures should be generated for a manuscript?

I am using tqdm and requests to manage file download in Python. However I can't figure out how to make tqdm display the progress bar in human-readable format i.e. in MB/s.

Here is my code

These are pytorch-yolo v3 code. I downloaded it in github. (https://github.com/eriklindernoren/PyTorch-YOLOv3) I tuned this for two classes. And while I'm doing trainning, there is still an error.

This is test.py code.