Image Segmentation on COCO-2017 Dataset

7PAM2015-0509-2024 -- Research Methods in Data Science

Mohit Agarwal (Student ID-22031257)

Project Overview

This project implements instance segmentation on the COCO-2017 dataset, focusing specifically on four classes: cake, car, person, and dog. The implementation uses Mask R-CNN with a ResNet-50 backbone and Feature Pyramid Network (FPN) for accurate object detection and instance segmentation.

• 📓 Colab Notebook

Dataset

The project uses a subset of the COCO-2017 dataset:

• Training set: 300 images with annotations
• Validation set: 300 images with annotations
• Test set: 30 images

The dataset is organized in COCO format with JSON annotation files containing information about object instances, categories, and segmentation masks.

Model Architecture

• Base Model: Mask R-CNN
• Backbone: ResNet-50 with Feature Pyramid Network (FPN)
• Target Classes: cake, car, person, dog

Project Structure

Assignment1/
├── coco_model_train.ipynb      # Main training and evaluation notebook
├── coco_eda.ipynb              # Main exploratory data analysis notebook
├── README.md                   # This file
├── RM_Segmentation_Assignment_dataset/
│   ├── train-300/              # Training dataset
│   │   ├── data/               # Training images
│   │   └── labels.json         # Training annotations
│   ├── validation-300/         # Validation dataset
│   │   ├── data/               # Validation images
│   │   └── labels.json         # Validation annotations
│   └── test-30/                # Test dataset
├── Images/                     # Directory for saving output images
└── MaskRCNN_Checkpoints/       # Directory for saving model checkpoints

Configuration

The model is trained with the following configuration:

• Batch size: 5
• Number of epochs: 10
• Learning rate: 0.0001
• LR scheduler step size: 3
• LR scheduler gamma: 0.1
• Device: CUDA (if available) or CPU
• Random seed: 42

Implementation Details

The implementation includes:

1. Custom COCO dataset loader to filter target classes
2. Mask R-CNN model with pre-trained weights
3. Training pipeline with optimizer and learning rate scheduler
4. Evaluation metrics including mean Average Precision (mAP)
5. Visualization tools for displaying predictions

Requirements

• ipykernel==6.29.5
• ipython==9.3.0
• jupyter_client==8.6.3
• jupyter_core==5.8.1
• matplotlib==3.10.3
• matplotlib-inline==0.1.7
• opencv-python==4.11.0.86
• pandas==2.3.0
• pycocotools==2.0.10
• seaborn==0.13.2
• torch==2.7.1
• torchaudio==2.7.1
• torchvision==0.22.1
• tqdm==4.67.1

Usage

1. Clone the repository
2. Set up the environment with required dependencies
3. Run the coco_model_train.ipynb notebook to train and evaluate the model
4. Run the coco_eda.ipynb notebook to perform exploratory data analysis on dataset

Results

The model is evaluated on the validation set using standard COCO metrics:

• mAP (mean Average Precision)
• AP at different IoU thresholds
• AP for different object sizes (small, medium, large)

Visualizations of the segmentation results are saved in the Images directory.

License

This project is part of the Research Methods in Data Science course assignment.

Acknowledgments

• COCO dataset creators
• PyTorch and torchvision teams
• Research Methods in Data Science course instructors