A Simple Framework for Contrastive Learning of Visual Representations

Open paper
Benchmark Model Rank Results
contrastive-learning-on-imagenet-1kResNet50#4ImageNet Top-1 Accuracy: 69.3
image-classification-on-places205SimCLR#13Top 1 Accuracy: 53.3
object-recognition-on-shape-biasSimCLR (ResNet-50x2)#11shape bias: 41.7
object-recognition-on-shape-biasSimCLR (ResNet-50x4)#12shape bias: 40.7
object-recognition-on-shape-biasSimCLR (ResNet-50x1)#13shape bias: 38.9
person-re-identification-on-sysu-30kSimCLR (self-supervised)#7Rank-1: 10.9
self-supervised-image-classification-on-1SimCLR (Resnet-50)#62Top 1 Accuracy: 77.2%
self-supervised-image-classification-on-imagenetSimCLR (ResNet-50 4x)#55Top 1 Accuracy: 76.5%Top 5 Accuracy: 93.2%Number of Params: 375M
self-supervised-image-classification-on-imagenetSimCLR (ResNet-50 2x)#79Top 1 Accuracy: 74.2%Top 5 Accuracy: 92.0%Number of Params: 94M
self-supervised-image-classification-on-imagenetSimCLR (ResNet-50)#96Top 1 Accuracy: 69.3%Top 5 Accuracy: 89.0%Number of Params: 24M
self-supervised-person-re-identification-onSimCLR#4Rank-1: 10.9
semi-supervised-image-classification-on-1SimCLR (ResNet-50 4×)#28Top 1 Accuracy: 63.0%Top 5 Accuracy: 85.8%
semi-supervised-image-classification-on-1SimCLR (ResNet-50 2×)#35Top 1 Accuracy: 58.5%Top 5 Accuracy: 83.0%
semi-supervised-image-classification-on-1SimCLR (ResNet-50)#46Top 1 Accuracy: 48.3%Top 5 Accuracy: 75.5%
semi-supervised-image-classification-on-2SimCLR (ResNet-50 4×)#43Top 5 Accuracy: 92.6%
semi-supervised-image-classification-on-2SimCLR (ResNet-50 2×)#45Top 5 Accuracy: 91.2%
semi-supervised-image-classification-on-2SimCLR (ResNet-50)#51Top 5 Accuracy: 87.8%

Rank counts only results with a code link.