如何在eval过程中获得train_reader同样的预处理

PaddlePaddle / PaddleDetection

Object Detection toolkit based on PaddlePaddle. It supports object detection, instance segmentation, multiple object tracking and real-time multi-person keypoint detection.

Apache License 2.0

12.8k stars 2.89k forks source link

如何在eval过程中获得train_reader同样的预处理 #4531

Closed ghost closed 3 years ago

ghost commented 3 years ago

最近在进行TTFNet的相关应用，今天想要在eval的过程中获取数据的预处理结果：ttf_heatmap、ttf_box_target，但是好像只是修改reader.yml文件中的EvalReader配置好像没有用：可能是evalreader不会去读取gt_box？我在trainer.py中看了半天希望将eval_reader修改为train_reader也无济于事，请问要怎么操作才能够在eval过程中获取预处理结果？

期待您的回复！

heavengate commented 3 years ago

configs/datasets目录下对应的数据集文件，参考TrainDataset配置一下data_fields https://github.com/PaddlePaddle/PaddleDetection/blob/d9187ad5d054bc8a4333b9a8ba686569be91a8c6/configs/datasets/coco_detection.yml#L9

ghost commented 3 years ago

configs/datasets目录下对应的数据集文件，参考TrainDataset配置一下data_fields

https://github.com/PaddlePaddle/PaddleDetection/blob/d9187ad5d054bc8a4333b9a8ba686569be91a8c6/configs/datasets/coco_detection.yml#L9

已解决！感谢您的回复！