PaddleOCR
PaddleOCR copied to clipboard
picodet_lcnet_x1_0_fgd_layout_table的推理模型的结果不是table类别
picodet_lcnet_x1_0_fgd_layout_table的推理模型(https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.7/ppstructure/docs/models_list.md)中,生成的结果类型为text类别,这个只是检测表格,我理解应该只有table类别才对呀
看下inference的yaml长啥样呢
infer_cfg.yml内容: mode: paddle draw_threshold: 0.5 metric: COCO use_dynamic_shape: false arch: PicoDet min_subgraph_size: 3 Preprocess:
- interp: 2
keep_ratio: false
target_size:
- 800
- 608 type: Resize
- is_scale: true
mean:
- 0.485
- 0.456
- 0.406 std:
- 0.229
- 0.224
- 0.225 type: NormalizeImage
- type: Permute
- stride: 32 type: PadStride label_list:
- table NMS: keep_top_k: 100 name: MultiClassNMS nms_threshold: 0.5 nms_top_k: 1000 score_threshold: 0.3 fpn_stride:
- 8
- 16
- 32
- 64
版面检测的结果,标签为text
在 2024-03-21 23:28:59,"cuicheng01" @.***> 写道:
看下inference的yaml长啥样呢
— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>
这边建议您使用COCO格式数据集重新标注一下,只留下table一个类别,并按照表格识别的文档进行训练