The Image Recommender 2023 is a powerful tool that identifies visually similar images by analyzing color and shape features. It integrates advanced big data analysis tools, including machine learning and vector similarity search.
- Utilizes a pre-trained MobileNet model from Keras for feature extraction.
- Computes and compares HSV and RGB color histograms.
- Employs Euclidean, Manhattan, and cosine distances for similarity calculation.
- Implements FAISS (Facebook AI Similarity Search) for efficient similar image retrieval.
This project is designed to work with a local dataset of images. The image collection used for development, comprising over 500,000 images, is not provided due to storage constraints.
- Feature Extraction
- Database Interaction
- FAISS Index Construction
- Similarity Calculation
- Similar Image Retrieval
Here is an example of the tool's output, displaying a test image and its top 5 similar images based on the algorithm:
