CeiT-S (384 finetune resolution)

Reported on 6 benchmarks across 1 task · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Computer Vision6 results

Image ClassificationonStanford Cars
Accuracy· 2021-03-22
94.1
best: 96.868 (efficient adaptive ensembling)
Incorporating Convolution Designs into Visual Transformers arXiv:2103.11816
Image ClassificationonCIFAR-10
Percentage correct· uses extra data· 2021-03-22
99.1
best: 99.5 (ViT-H/14)
Incorporating Convolution Designs into Visual Transformers arXiv:2103.11816
Image ClassificationonOxford-IIIT Pets
Accuracy· 2021-03-22
94.9
best: 97.1 (EffNet-L2 (SAM))
Incorporating Convolution Designs into Visual Transformers arXiv:2103.11816
Image ClassificationonFlowers-102
Accuracy· uses extra data· 2021-03-22
98.6
best: 99.76 (CCT-14/7x2)
Incorporating Convolution Designs into Visual Transformers arXiv:2103.11816
Image ClassificationoniNaturalist 2019
Top-1 Accuracy· uses extra data· 2021-03-22
82.7
best: 88.5 (Hiera-H (448px))
Incorporating Convolution Designs into Visual Transformers arXiv:2103.11816
Image ClassificationonCIFAR-100
Percentage correct· uses extra data· 2021-03-22
91.8
best: 96.08 (EffNet-L2 (SAM))
Incorporating Convolution Designs into Visual Transformers arXiv:2103.11816