Image classification
July 6, 2018 ยท View on GitHub
Description
Assigning a single class label to the whole input image.
Metrics:
- Top-1 Accuracy
- Top-5 Accuracy
Results
ImageNet
Single model, single crop. Top-5 results by Acc@5 are highlighted in boldface as well.
| Model | Acc@1 | Acc@5 | Weights | Paper | Code |
|---|---|---|---|---|---|
| DenseNet-121 | 82.858 | 92.290 | PyTorch | paper | PyTorch |
| DenseNet-161 | 82.736 | 93.798 | PyTorch | paper | PyTorch |
| DenseNet-169 | 82.693 | 93.150 | PyTorch | paper | PyTorch |
| DenseNet-201 | 82.566 | 93.660 | PyTorch | paper | PyTorch |
| Inception-v1 | 81.320 | 89.600 | TensorFlow | paper | TensorFlow |
| Inception-v2 | 81.304 | 91.800 | TensorFlow | paper | TensorFlow |
| Inception-v3 | 80.400 | 93.900 | TensorFlow PyTorch | paper | TensorFlow PyTorch |
| Inception-v4 | 80.200 | 95.200 | TensorFlow | paper | TensorFlow |
| Inception-v4 | 80.170 | 94.930 | PyTorch | paper | PyTorch |
| Inception-ResNet-v2 | 80.060 | 95.300 | TensorFlow | paper | TensorFlow |
| Inception-ResNet-v2 | 78.888 | 95.234 | PyTorch | paper | PyTorch |
| NASNet-A-Large | 78.428 | 96.163 | TensorFlow | paper | TensorFlow |
| NASNet-A-Large | 78.000 | 96.086 | PyTorch | paper | PyTorch |
| PNASNet-5-Large | 77.560 | 96.182 | TensorFlow | paper | TensorFlow |
| PNASNet-5-Large | 77.438 | 95.992 | PyTorch | paper | PyTorch |
| SENet154 | 77.420 | 95.530 | Caffe | paper | Caffe |
| FBResNet-152 | 77.386 | 93.594 | PyTorch | paper | PyTorch |
| SENet154 | 76.200 | 95.498 | PyTorch | paper | PyTorch |
| Xception | 74.980 | 94.292 | PyTorch | paper | PyTorch |
| ResNet-101 | 73.900 | 93.672 | PyTorch | paper | PyTorch |
PASCAL VOC Classification
TODO: add dataset description
| Model | Score | Weights | Paper | Code |
|---|---|---|---|---|
| Model name | 0.5 | link to the weights of a trained model | paper | link to implementation |
Resources
| Title | Type | Description |
|---|---|---|
| pretrained-models.pytorch | Repository | Repo dedicated to reproduction of research paper results and collect pretrained convolutional nets for classification (PyTorch). |
| tf/research/slim | Repository | Convolutional nets for classification pretrained on ImageNet (TF Slim) |