YOLOv5 custom dataset (Easy Guide)

Please use this on google colab environment. Also, the provided notebook is using mask detector as the example dataset
You can find the available dataset at here
This example is using the URL download link, which will download the data into temporary repository in google colab (won't consume google drive space)

Three main steps and you're good to go

• Setup Environment
• Training Process
• Testing the model

Setup Environment

• Create temporary yolov5 repository that is cloned from here
• Import dependencies

Training Process

• Obtain data from open source . This example is using URL to download data in form of images and labels
  
• Check and change directory for training
• Train the YOLOv5s algorith with data.yaml and weights
  Note: You can use pretrained weights for any YOLOv5 architecture or just use weights = "" for randomized initial weights

Testing the model

Just deploy using the detect.py provided in YOLOv5 repo
Remember to provide:

1. source (the images or video you want to test with)
2. data (the data.yaml for the classes)
3. weights (the trained model weight, we use best.pt in this example)