Instructions for Running MATLAB Files and YOLOv5 Training/Detection

This repository contains MATLAB scripts for circle detection, distance calculation, image splitting, and a Python script for YOLOv5 model training and cell detection.

Requirements

• MATLAB installed on your system.
• Python environment with required packages for YOLOv5.

MATLAB Scripts

File Description

• CircleDetectionAndRadius.m: Detects circles in an image, saves the detected circles as separate files, and computes their radii and coordinates.
• Location and Distance.m: Detects circles in an image, saves the detected circles as separate files, calculates their radii and coordinates, and computes distances between circles.
• SecondRound.m: Splits input images into smaller patches with specified patch size and overlap.

Instructions

1. CircleDetectionAndRadius.m:

   • Open MATLAB.
   • Ensure the image file (Linear_1.png or Linear_1.tif) is in the same directory as the script.
   • Modify parameters such as radiusRange, sensitivity, and edgeThreshold if needed.
   • Run the script. Detected circles will be saved in the Linear_1 folder.

2. Location and Distance.m:

   • Follow the same steps as above but with a different image file (Linear_1.tif).
   • Detected circles and their centers will be saved in the Linear_1 folder.
   • Distances from a randomly chosen circle to other circles will be calculated and saved.

3. SecondRound.m:

   • This script requires a set of input images (Linear_1.png, Linear_2.png, etc.) in the same directory.
   • Modify patchSize and overlap variables as needed.
   • Run the script. Cropped patches will be saved in the images folder.

YOLOv5 Training and Detection

File Description

• YOLODLModel.py: Python script for training YOLOv5 on custom data for 100 epochs and detecting cells.

Instructions for YOLOv5

1. YOLODLModel.py:

   • Ensure Python is installed on your system along with required YOLOv5 dependencies.
   • Place the YOLODLModel.py file in a desired directory.
   • Modify parameters like --img, --batch, --epochs, etc., as per your requirements in the script.
   • Execute the script in a terminal or command prompt:

     python YOLODLModel.py

   • The script will start training the YOLOv5 model on custom data we provided. It will generate output files and folders based on specified parameters.

2. Evaluation and Cell Detection:

   • After training, for model evaluation and cell detection, use the provided code snippet in a Python environment or Jupyter Notebook:

     from utils.plots import plot_results
     from IPython.display import Image
     
     # Plot results.txt as results.png
     plot_results('/content/yolov5/runs/train/yolov5s_results2/results.png', width=1000)
     
     # View results.png
     Image(filename='/content/yolov5/runs/train/yolov5s_results2/results.png', width=1000)