OverFeat
A convolutional network-based image classifier and feature extractor trained on ImageNet, providing dense feature extraction capabilities.
What is OverFeat?
OverFeat is a Convolutional Neural Network-based system for image classification and dense feature extraction. It was trained on the ImageNet dataset and provides tools to recognize images and extract generic visual features that can be used for various computer vision tasks. The project includes a library with C++ source code and wrappers for scripting languages like Python and Lua.
Researchers and developers in computer vision who need a pre-trained CNN for image classification or a generic feature extractor for transfer learning and vision experiments.
OverFeat offers a state-of-the-art, integrated approach to recognition, localization, and detection with efficient dense feature extraction. It provides pre-trained models, multiple network sizes, and extensive APIs, making it a versatile tool for vision research without requiring training from scratch.
Overview
OverFeat is a Convolutional Neural Network (CNN) designed for image classification and dense feature extraction. It was trained on the ImageNet dataset and participated in the ImageNet 2013 competition, serving as a foundational tool for researchers in computer vision.
Key Features
- Image Classification — Recognizes images and outputs top class predictions with configurable result counts.
- Feature Extraction — Extracts dense feature vectors from images, useful for downstream vision tasks.
- Multi-Scale Networks — Supports both small (faster) and large (more accurate) network architectures.
- Webcam Demo — Includes a real-time classification demo using webcam input.
- Batch Processing — Processes entire directories of images for feature extraction or classification.
- Layer Output Access — Allows extraction of features from any intermediate layer of the network.
Philosophy
OverFeat emphasizes integrated recognition, localization, and detection through a unified convolutional network architecture, aiming to provide a generic and efficient feature extractor for vision research.
Related Projects
Found a gem we're missing?
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
