In this repo, I've trained an object detection model on a custom dataset to find the number of RBC, WBC, PLATELETS Count from the microscopic blood-smeared images. Original Dataset available at here.
Preprocessed dataset for Yolo-v5 available at here
Built using Yolo-v5 model open-sourced by ultralytics, more at here
I also wrote a medium post describing the whole process, If interested check it out - medium
Simply start the training and validation here and to begin inference
The Inference notebook is similar to move the model in production environment. You need to these files in your system path for proper imports, you can get them from here Yolo-v5
Excerpts :
and