0

PaddleDetection:deploy/python/infer.py的batch_size设置>1遍报错 error message：

Traceback (most recent call last):
File "/home/zyp/program/pdf_ocr/PaddleDetection-release-2.6/deploy/python/infer.py", line 1097, in
    main()
File "/home/zyp/program/pdf_ocr/PaddleDetection-release-2.6/deploy/python/infer.py", line 1063, in main
    save_results=FLAGS.save_results)
File "/home/zyp/program/pdf_ocr/PaddleDetection-release-2.6/deploy/python/infer.py", line 413, in predict_image
    result = self.postprocess(inputs, result)
File "/home/zyp/program/pdf_ocr/PaddleDetection-release-2.6/deploy/python/infer.py", line 682, in postprocess
    np_boxes, np_boxes_num = postprocessor(np_score_list, np_boxes_list)
File "/home/zyp/program/pdf_ocr/PaddleDetection-release-2.6/deploy/python/picodet_postprocess.py", line 166, in call
    center = center[topk_idx]
IndexError: index 165237 is out of bounds for axis 0 with size 7600

model：picodet_lcnet_x1_0_fgd_layout_cdla_infer

I tried many times, it works normally when batch_size=1, but when batch_size > 1, the above error is reported, does anyone know why?

toyota Supra
  • 3,181
  • 4
  • 15
  • 19
user21860229
  • 1
  • In the documentation of paddledetection, I found the following information: Note: The Paddle prediction library version needs to be >=2.1, and batch_size>1 only supports YOLOv3 and PP-YOLO. The model I use does not support batch_size>1 – user21860229 Aug 10 '23 at 02:35

0 Answers0