#382 · Primary category: Computer Vision

miles-deep

Deep Learning Porn Video Classifier/Editor with Caffe

Project last updated:08/23/18

GitHub Stars

2.7K

Forks

279

Contributors

1

License

GPL-3.0

Why we included this project

Miles Deep tackles a problem most NSFW detectors ignore: they flag nudity, while this tool tells several specific sexual acts apart and then edits the footage based on that. It runs a convolutional network with residual connections across each second of video, sorts frames into six categories, and lets you strip out everything that isn't the act you're looking for. Both the analysis and the automated cutting live in one small command-line tool. Developers building content-moderation or custom video-classification pipelines will find the C++ batching and threading setup worth studying, since it keeps throughput high on a mid-range GPU. The catch is the stack: it expects Caffe with CUDA 8, so getting it running on a current machine takes some setup work.

Articles for this project

No articles for this project yet.

To suggest a topic or contribute an article, contact us.

Related projects in this category