Box2Poly: Memory-Efficient Polygon Prediction of Arbitrarily Shaped and Rotated Text
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2023
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866908484552884224 |
|---|---|
| author | Chen, Xuyang Wang, Dong Schindler, Konrad Sun, Mingwei Wang, Yongliang Savioli, Nicolo Meng, Liqiu |
| author_facet | Chen, Xuyang Wang, Dong Schindler, Konrad Sun, Mingwei Wang, Yongliang Savioli, Nicolo Meng, Liqiu |
| contents | Recently, Transformer-based text detection techniques have sought to predict polygons by encoding the coordinates of individual boundary vertices using distinct query features. However, this approach incurs a significant memory overhead and struggles to effectively capture the intricate relationships between vertices belonging to the same instance. Consequently, irregular text layouts often lead to the prediction of outlined vertices, diminishing the quality of results. To address these challenges, we present an innovative approach rooted in Sparse R-CNN: a cascade decoding pipeline for polygon prediction. Our method ensures precision by iteratively refining polygon predictions, considering both the scale and location of preceding results. Leveraging this stabilized regression pipeline, even employing just a single feature vector to guide polygon instance regression yields promising detection results. Simultaneously, the leverage of instance-level feature proposal substantially enhances memory efficiency (>50% less vs. the state-of-the-art method DPText-DETR) and reduces inference speed (>40% less vs. DPText-DETR) with minor performance drop on benchmarks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2309_11248 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Box2Poly: Memory-Efficient Polygon Prediction of Arbitrarily Shaped and Rotated Text Chen, Xuyang Wang, Dong Schindler, Konrad Sun, Mingwei Wang, Yongliang Savioli, Nicolo Meng, Liqiu Computer Vision and Pattern Recognition Recently, Transformer-based text detection techniques have sought to predict polygons by encoding the coordinates of individual boundary vertices using distinct query features. However, this approach incurs a significant memory overhead and struggles to effectively capture the intricate relationships between vertices belonging to the same instance. Consequently, irregular text layouts often lead to the prediction of outlined vertices, diminishing the quality of results. To address these challenges, we present an innovative approach rooted in Sparse R-CNN: a cascade decoding pipeline for polygon prediction. Our method ensures precision by iteratively refining polygon predictions, considering both the scale and location of preceding results. Leveraging this stabilized regression pipeline, even employing just a single feature vector to guide polygon instance regression yields promising detection results. Simultaneously, the leverage of instance-level feature proposal substantially enhances memory efficiency (>50% less vs. the state-of-the-art method DPText-DETR) and reduces inference speed (>40% less vs. DPText-DETR) with minor performance drop on benchmarks. |
| title | Box2Poly: Memory-Efficient Polygon Prediction of Arbitrarily Shaped and Rotated Text |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2309.11248 |