Abstract: This project introduces ElephantMatcher, a machine learning method to detect, extract features, and identify elephants in both aerial and ground-based wildlife images. The system is suitable for ecological applications such as re-identification and population monitoring , contributing to conservation efforts through automated processes. The method uses the YOLOv8 object detection model to detect elephant regions from high-resolution images. These regions are then preprocessed through grayscale conversion, resizing, and Canny edge detection to enhance feature extraction. Feature extraction is performed using Histogram of Oriented Gradients (HOG) for capturing shape-related information and Local Binary Patterns (LBP) for texture analysis. The extracted features are combined into a multi-descriptor signature used for similarity comparison. Identification is achieved through cosine similarity measurements between feature signatures. ElephantMatcher offers a practical foundation for scalable wildlife monitoring systems that support long-term elephant conservation.