How to use it
- Open the collection and pick your framework.
- Find a model for your task: classification, detection, segmentation and more.
- Follow the link, load the weights and fine-tune on your data.
How it's built
A curated, categorised markdown index of computer-vision pretrained models across TensorFlow, PyTorch, Keras, Caffe and MXNet, with short descriptions and links.
Why it matters for businesses
Training vision models from scratch costs weeks of compute and expertise. Starting from a proven pretrained model cuts both, so features ship sooner.
In real life
A team building a defect-detection system for a factory starts from a pretrained detector and fine-tunes it on a few hundred labelled photos.



