developer0hye/PyTorch-Deformable-Convolution-v2
Don't feel pain to use Deformable Convolution
This project helps deep learning engineers improve the accuracy of image classification models, especially when dealing with variations in object scale or pose. By integrating Deformable Convolution v2 (DCNv2) layers into a neural network, it enables models to better adapt to geometric transformations in input images. The engineer adds this component to their PyTorch model, and the output is a more robust classifier that performs better on scaled or distorted images.
343 stars. No commits in the last 6 months.
Use this if you are a deep learning engineer building image classification models in PyTorch and need to improve their performance on images with significant scale variations or other geometric distortions, without extensive data augmentation.
Not ideal if you are working with non-image data, or if you need to convert your model to ONNX format, as that is not yet supported.
Stars
343
Forks
33
Language
Jupyter Notebook
License
MIT
Category
Last pushed
Jun 07, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/developer0hye/PyTorch-Deformable-Convolution-v2"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
deepinv/deepinv
DeepInverse: a PyTorch library for solving imaging inverse problems using deep learning
fidler-lab/polyrnn-pp
Inference Code for Polygon-RNN++ (CVPR 2018)
mhamilton723/STEGO
Unsupervised Semantic Segmentation by Distilling Feature Correspondences
yjxiong/tsn-pytorch
Temporal Segment Networks (TSN) in PyTorch
pyxu-org/pyxu
Modular and scalable computational imaging in Python with GPU/out-of-core computing.