TResNet: High Performance GPU-Dedicated Architecture
A PyTorch implementation of TResNet, a high-performance convolutional neural network architecture optimized for GPU training and inference.
What is TResNet: High Performance GPU-Dedicated Architecture?
TResNet is a family of convolutional neural network architectures designed to deliver superior accuracy while maintaining high GPU training and inference throughput. It addresses the limitations of FLOPs-optimized models by introducing modifications that better utilize GPU structure, offering a better speed-accuracy trade-off than many contemporary networks.
Machine learning researchers and engineers who need efficient, high-accuracy convolutional neural networks for computer vision tasks, particularly those working with GPU-accelerated training and inference pipelines.
Developers choose TResNet because it prioritizes actual GPU efficiency over theoretical FLOPs count, achieving state-of-the-art accuracy on ImageNet and transfer learning datasets while offering higher images-per-second throughput than ResNet50 and other networks with similar accuracy.
Overview
Official Pytorch Implementation of "TResNet: High-Performance GPU-Dedicated Architecture" (WACV 2021)
Use Cases
Best For
- Achieving state-of-the-art accuracy on ImageNet classification with high GPU throughput (e.g., 80.7% top-1 accuracy with TResNet-M).
Related Projects
Found a gem we're missing?
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
