Object Detection and Categorization for Image and Video

Problem Statement

Imagine effortlessly analysing and categorising vast collections of images and videos in a snap! With the power of computer vision, we can develop a cutting-edge model that accurately detects and categorizes different objects, providing us with invaluable insights and saving us hours of tedious manual work. From identifying objects' presence and location to classifying them into relevant categories, our model will revolutionize how we approach image and video analysis. Say goodbye to the hassle of sifting through endless files, and say hello to the future of efficient and automated image and video processing!

OUTPUT:

Images are stored in a folder without segregation.

Sorted images in specific folders.

Dataset

Download Dataset: (https://github.com/MargiPandya27/Intel-Hackathon-Image-to-Folder/blob/main/archive_(5).zip)

Pre-Requisites(prefered)

Softwares:

1. OpenAPI,
2. Google Colab/Jupyter Notebook

Packages:

1. Python >= 3.5
2. Torch >= 1.8.1
3. Torchvision >= 0.9.1

NOTE: Make sure that dataset is as specified. Put the .yaml file into yolov5 folder after cloning it. Change the location in the avenger.yaml according to your dataset location.

Training on custom dataset

1. Form a dataset folder
2. Make a folder named images and label
3. Upload the files.
4. Perform Augmentation using Augmentation.ipynb
5. Define paths and class labels in .yaml file.

References:

Augmentation:

https://www.learnpytorch.io/04_pytorch_custom_datasets/

Object Detection

https://github.com/nicknochnack/YOLO-Drowsiness-Detection