ahelou2/deep-learning-scene-recognition
Using convolutional deep belief nets to learn natural scene representations
This repo contains code for training a deep learning model on a set of unlabled images in order to extract invariant features. Such invariant features can then be used for image recognition with a generic machine learning algorithm.
Public repository record indexed from GitHub. Explore verified star velocity metrics, source code repositories, and curated developer tool directories across the GitHubRepo ecosystem.
Using convolutional deep belief nets to learn natural scene representations