DAVAR-Lab-OCR
DAVAR-Lab-OCR copied to clipboard
hikopensource
OCR toolbox from Davar-Lab
Hello, when I was downloading the model docbank.pth, it was displayed that the current Hikvision network disk (one. hikvision. com) has been deactivated. Please migrate to the HikLink cloud disk...
The inference function 'simple_test' seems only support Multi-Modal Feature Fusion part and Relational Graph Construction part, can you open source the Table Elements Extraction part and Structural Graph Construction part?...
我在可视化label的过程发现，部分图片的标注有误，标注信息如下： "texts": ["如何选择合适尺寸的卷发棒", "多尺寸满足更多需求", "俏皮内扣，空气刘海，", "唯美梨花，气质中卷，波浪大卷", "*锁骨发建议", "*中长发建议", "*及腰长发建议", "选用26mm规格", "选用32mm规格", "选用38mm规格"], "labels": [[1],[2],[3],[4],[5],[7],[9],[6],[8],[10]] 这是阅读顺序的可视化图片： 
| Models | Precision | Recall | Hmean | Links | |----------------|-----------|---------|----------|-------| | Text Detection | 99.57 | 99.53 | 99.55 | [config](https://github.com/hikopensource/DAVAR-Lab-OCR/blob/main/demo/table_recognition/lgpma/tools/ocr_models/det_mask_rcnn_r50_fpn_pubtabnet.py), [pth](https://drive.hikvision.com/hcs/controller/hik-manage/fileDownload?link=ZtMP7Cok) (Access Code: N243) | | Models...
["", "", "", "", "", "", "", "", "", "", "", "", "", "", "", "", "", "", "", "", "", "", "", "", "", "", "", "", "", "",...
训练训练自己的数据报错： File "/root/miniconda/envs/d2/lib/python3.7/site-packages/mmdet/models/detectors/base.py", line 181, in forward return self.forward_train(img, img_metas, **kwargs) File "/root/workspace/DAVAR-Lab-OCR-main/davarocr/davar_table/models/detectors/lgpma.py", line 134, in forward_train **kwargs) File "/root/workspace/DAVAR-Lab-OCR-main/davarocr/davar_table/models/roi_heads/lgpma_roi_head.py", line 86, in forward_train gt_masks, img_metas, gt_bboxes) File "/root/workspace/DAVAR-Lab-OCR-main/davarocr/davar_table/models/roi_heads/lgpma_roi_head.py", line...
不明白第一轮正常跑完，到第三报错了，看报错原因应该是数据的问题 2023-03-22 20:52:55,756 - davarocr - INFO - Epoch [3][2610/8424] lr: 1.000e-02, eta: 1 day, 11:59:42, time: 2.943, data_time: 0.423, memory: 15020, loss_rpn_cls: 0.0664, loss_rpn_bbox: 0.0797, loss_cls: 0.1295, acc: 94.2090,...
