ASTER

Reported on 9 benchmarks across 3 tasks

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

Computer Vision6 results

Scene ParsingonSVT
Accuracy
89.5
best: 99.1 (CLIP4STR-H (DFN-5B))
Scene ParsingonICDAR2015
Accuracy
76.1
best: 93.5 (DTrOCR 105M)
Scene ParsingonICDAR2013
Accuracy
91.8
best: 99.42 (CLIP4STR-L*)
Scene Text RecognitiononSVT
Accuracy
89.5
best: 99.1 (CLIP4STR-H (DFN-5B))
Scene Text RecognitiononICDAR2015
Accuracy
76.1
best: 93.5 (DTrOCR 105M)
Scene Text RecognitiononICDAR2013
Accuracy
91.8
best: 99.42 (CLIP4STR-L*)

Audio3 results

2D Semantic SegmentationonSVT
Accuracy
89.5
best: 99.1 (CLIP4STR-H (DFN-5B))
2D Semantic SegmentationonICDAR2015
Accuracy
76.1
best: 93.5 (DTrOCR 105M)
2D Semantic SegmentationonICDAR2013
Accuracy
91.8
best: 99.42 (CLIP4STR-L*)