Training a TensorFlow Yolov5 Object Detection Model on enhanced BCCD

Introduciton

One-stage object detection model Faster-rcnn were applied to classify

neutrophils, eosinophils, monocytes, and lymphocytes on an enhanced BCCD dataset.

Preparing Images and Annotations

BCCD image is avaliable on https://www.kaggle.com/paultimothymooney/blood-cells

Image annotation can accomplish by labelImg https://github.com-/tzutalin/labelImg

Creating Yolo.format annotation files

Transform your annotation file to .txt file.

The whole Yolov5 Dataset Structure is as follows:

- bcc
  - Images
      - Train (.jpg files)
      - Valid (.jpg files)
  - labels
      - Train (.txt files)
      - Valid (.txt files)

Training the model on GoogleColab

Inference on unseen images