Model card
TextBPN++ (ResNet-50+DCN).
Zhang et al.open-sourceUnknown paramsResNet-50 with Deformable Convolution + Boundary Transformer
TextBPN++ best variant with deformable convolution backbone. arXiv:2205.05320. T-MM 2023.
§ 01 · Benchmarks
Every benchmark TextBPN++ (ResNet-50+DCN) has a recorded score for.
| # | Benchmark | Area · Task | Metric | Value | Rank | Date | Source |
|---|---|---|---|---|---|---|---|
| 01 | msra-td500 | Computer Vision · Scene Text Detection | f-measure | 90.1% | #2 | 2022-05-11 | source ↗ |
| 02 | msra-td500 | Computer Vision · Scene Text Detection | precision | 93.7% | #2 | 2022-05-11 | source ↗ |
| 03 | icdar-2017-mlt | Computer Vision · Scene Text Detection | f-measure | 77.5% | #3 | 2022-05-11 | source ↗ |
| 04 | icdar-2017-mlt | Computer Vision · Scene Text Detection | precision | 83.7% | #3 | 2022-05-11 | source ↗ |
| 05 | icdar-2017-mlt | Computer Vision · Scene Text Detection | recall | 72.1% | #3 | 2022-05-11 | source ↗ |
| 06 | msra-td500 | Computer Vision · Scene Text Detection | recall | 86.8% | #4 | 2022-05-11 | source ↗ |
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 02 · Strengths by area
Where TextBPN++ (ResNet-50+DCN) actually performs.
§ 03 · Papers
1 paper with results for TextBPN++ (ResNet-50+DCN).
- 2022-05-11· Computer Vision· 6 results
Arbitrary Shape Text Detection via Boundary Transformer
§ 04 · Related models
Other Zhang et al. models scored on Codesota.
§ 05 · Sources & freshness
Where these numbers come from.
arxiv
6
results
6 of 6 rows marked verified.