AnimalsRCNN - Animal Detection and Classification with Mask R-CNN

Overview

This repository contains an implementation of the Mask R-CNN algorithm for detecting and classifying animals in images. Mask R-CNN is an advanced deep learning model that not only identifies objects within an image but also provides pixel-level segmentation, making it ideal for tasks that require precise object delineation.

The project aims to develop a robust system for wildlife monitoring and research, using deep learning to analyze images captured in natural habitats.

Table of Contents

• Features
• Dataset
• Installation
• Usage
• Results
• Technologies
• Contributing
• License
• Contact

Features

• Object Detection: Detects animals in images with bounding boxes.
• Instance Segmentation: Provides pixel-level masks for detected animals.
• Classification: Classifies detected animals into predefined categories.
• Model Training: Includes scripts to train the Mask R-CNN model on custom datasets.
• Evaluation: Tools to evaluate model performance on test datasets.

Dataset

The model is trained on a custom dataset containing images of various animals. Each image is annotated with bounding boxes, segmentation masks, and class labels.

• Data Format: The dataset should be in COCO format or similar, with separate files for training and validation sets.
• Classes: The dataset includes multiple animal classes, such as elephants, lions, and zebras.

Installation

Clone the repository and navigate to the project directory:

git clone https://github.com/YushenSun/AnimalsRCNN.git
cd AnimalsRCNN

Ensure you have Python and the required libraries installed:

pip install -r requirements.txt

The project requires the installation of additional dependencies, such as TensorFlow, Keras, and OpenCV. You can install them with:

pip install tensorflow keras opencv-python

Usage

Training the Model

To train the Mask R-CNN model on your dataset:

1. Place your dataset in the datasets/ directory.

2. Update the configuration file with the appropriate dataset path and parameters.

3. Run the training script:

   python train.py --dataset=path_to_your_dataset --epochs=50

Inference

To perform inference on new images:

1. Place the images in the images/ directory.

2. Run the inference script:

   python inference.py --image=path_to_image

The script will output the image with detected animals, along with their bounding boxes and masks.

Evaluation

To evaluate the model performance on a test set:

python evaluate.py --dataset=path_to_test_dataset

This script will output metrics such as mean Average Precision (mAP) and segmentation accuracy.

Results

The model achieves high accuracy in detecting and classifying animals across various test images. Below are some example results:

• Elephant Detection: 95% accuracy with clear segmentation.
• Lion Detection: 92% accuracy with well-defined bounding boxes.
• Zebra Detection: 94% accuracy with accurate masks.

Sample images from the test set, along with their corresponding detection and segmentation results, can be found in the results/ directory.

Technologies

• Deep Learning: TensorFlow, Keras
• Computer Vision: OpenCV
• Instance Segmentation: Mask R-CNN
• Data Handling: COCO format, JSON
• Visualization: Matplotlib, Seaborn

Contributing

Contributions are welcome! If you want to improve the model, add new features, or fix bugs, please follow these steps:

1. Fork the repository.
2. Create a new branch.
3. Make your changes and commit them.
4. Submit a pull request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contact

For any questions, suggestions, or feedback, feel free to contact me:

• Yushen Sun
• LinkedIn
• Email