WalterSimoncini/fungivision
Library implementation of "No Train, all Gain: Self-Supervised Gradients Improve Deep Frozen Representations"
This library helps machine learning practitioners improve the performance of image, text, or audio classification and image segmentation tasks. It takes raw input data (like images from a dataset) and a pre-trained vision transformer model. It then generates enhanced "gradient features" that, when combined with the model's standard outputs, lead to more accurate downstream predictions. This is for researchers or engineers working with self-supervised learning for feature extraction.
No commits in the last 6 months.
Use this if you are using pre-trained vision transformer models and want to boost their performance on tasks like image classification or semantic segmentation without extensive retraining.
Not ideal if you need a solution for models other than vision transformers or if your primary goal is to train models from scratch, as this focuses on improving frozen representations.
Stars
40
Forks
—
Language
Python
License
MIT
Category
Last pushed
Oct 31, 2024
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/WalterSimoncini/fungivision"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
Jittor/jittor
Jittor is a high-performance deep learning framework based on JIT compiling and meta-operators.
zhanghang1989/ResNeSt
ResNeSt: Split-Attention Networks
berniwal/swin-transformer-pytorch
Implementation of the Swin Transformer in PyTorch.
NVlabs/FasterViT
[ICLR 2024] Official PyTorch implementation of FasterViT: Fast Vision Transformers with...
ViTAE-Transformer/ViTPose
The official repo for [NeurIPS'22] "ViTPose: Simple Vision Transformer Baselines for Human Pose...