在使用ocr时遇到crop by polys 问题

[TimothyZero]

🔎 Search before asking

• [x] I have searched the PaddleOCR Docs and found no similar bug report.
• [x] I have searched the PaddleOCR Issues and found no similar bug report.
• [x] I have searched the PaddleOCR Discussions and found no similar bug report.

🐛 Bug (问题描述)

Paddlex的crop by polys会在sample points on Bbox 时生成＜三个点的points，最终导致ocr crop失败

🏃‍♂️ Environment (运行环境)

都是3.0.0

🌰 Minimal Reproducible Example (最小可复现问题的Demo)

sample_points_on_bbox(np.array([[436, 188], [432, 188], [432, 191], [427, 194], [424, 194], [416, 188], [403, 188], [399, 185], [399, 183], [398, 181], [396, 178], [399, 177], [436, 177], [439, 180], [439, 185]]))

[cuicheng01]

您好，可以稍微再描述细致一些吗？

[TimothyZero]

就是我在做ocr时，text_type设置为seal时会使用CropByPolys剪裁文本行，然后在把文本行送去rec的到最终识别结果。但是问题是CropByPolys.sample_points_on_bbox会对polys points进行降采样导致一些正常的多边形降为3个点的，然后3个点无法进行剪裁会报错assert len(points) == 4, "shape of points must be 4*2"

[Sunting78]

您好，方便提供一下代码和图像吗？我们复现排查下

[TimothyZero]

图和代码不太方便，但是你可以直接跑这个样例，sample_points_on_bbox(np.array([[436, 188], [432, 188], [432, 191], [427, 194], [424, 194], [416, 188], [403, 188], [399, 185], [399, 183], [398, 181], [396, 178], [399, 177], [436, 177], [439, 180], [439, 185]]))，这个代码应保证输出至少4个点