RebanMark/Animal-Identification-using-RESNET50

An object detection project using Faster R-CNN (ResNet50 FPN) to classify and localize African wildlife species from images. Built with PyTorch and trained on the African Wildlife dataset from Kaggle. # 🦁 African Wildlife Detection with Faster R-CNN This project implements an object detection system to classify and localize African wildlife species using the **Faster R-CNN** model with a **ResNet50 FPN** backbone. Built with PyTorch, the system was trained and evaluated on the African Wildlife Dataset. --- ## 📌 Project Objective To accurately identify and locate animals in wildlife images, helping support efforts in conservation, monitoring, and automated analysis of African fauna. --- ## 📁 Dataset - **Source**: Kaggle - African Wildlife Dataset - **Contents**: High-quality images of wildlife species with bounding box annotations for: - Elephant - Zebra - Lion - Giraffe - Buffalo - Rhino --- ## 🛠️ Tools & Technologies - **Language**: Python - **Libraries**: PyTorch, Torchvision, Pandas, NumPy, PIL, Matplotlib - **Model**: Faster R-CNN with ResNet-50 FPN (Feature Pyramid Network) - **Notebook**: `reban-miniproject.ipynb` --- ## 🔍 Features - Custom dataset loader for bounding box annotations - Image augmentation and transformation pipeline - Transfer learning with pretrained Faster R-CNN - Visualization of predictions with bounding boxes - Train/validation/test splitting and evaluation --- ## 🖼️ Sample Prediction --- ## 🚀 How to Run 1. Clone this repository 2. Download the dataset from Kaggle and place it in your working directory 3. Open `Animal Identification.ipynb` in Jupyter or Colab 4. Run all cells to train and evaluate the model --- ## 📈 Results The model demonstrates strong performance in recognizing and localizing multiple animals in diverse scenarios, even with limited data. --- ## 🙋‍♂️ Author **Reban Mark** 📍 Coimbatore, India 📫 rebanmark1234@gmail.com --- ## ⭐ Acknowledgements - Bianca Ferreira for the African Wildlife dataset - PyTorch and Torchvision teams for the robust model support --- ## 🐾 Contributions Feel free to fork the repo, raise issues, and contribute improvements or new features!