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
12.38k
stars
2.84k
forks
source link
Ambiguity weight backbone.res5.res5a.branch2a.norm._mean loaded, it matches at least bbox_head.head.res5.res5a.branch2a.norm._mean and backbone.res5.res5a.branch2a.norm._mean in the model #7331
W1114 23:07:29.199206 18108 gpu_resources.cc:61] Please NOTE: device: 0, GPU Compute Capability: 8.6, Driver API Version: 11.6, Runtime API Version: 11.6
W1114 23:07:30.581168 18108 gpu_resources.cc:91] device: 0, cuDNN Version: 8.2.
[11/14 23:07:31] ppdet.utils.checkpoint INFO: The shape [512, 256, 3, 3] in pretrained weight backbone.res5.res5a.branch2a.conv.weight is unmatched with the shape [512, 1024, 1, 1] in model bbox_head.head.res5.res5a.branch2a.conv.weight. And the weight backbone.res5.res5a.branch2a.conv.weight will not be loaded
Traceback (most recent call last):
File "tools/train.py", line 172, in <module>
main()
File "tools/train.py", line 168, in main
run(FLAGS, cfg)
File "tools/train.py", line 129, in run
trainer.load_weights(cfg.pretrain_weights)
File "D:\0SDXX\PaddleDetection\ppdet\engine\trainer.py", line 375, in load_weights
load_pretrain_weight(self.model, weights)
File "D:\0SDXX\PaddleDetection\ppdet\utils\checkpoint.py", line 215, in load_pretrain_weight
param_state_dict = match_state_dict(model_dict, param_state_dict)
File "D:\0SDXX\PaddleDetection\ppdet\utils\checkpoint.py", line 194, in match_state_dict
weight_key]))
ValueError: Ambiguity weight backbone.res5.res5a.branch2a.norm._mean loaded, it matches at least bbox_head.head.res5.res5a.branch2a.norm._mean and backbone.res5.res5a.branch2a.norm._mean in the model
[X] 我确认已经提供了Bug复现步骤、代码改动说明、以及环境信息,确认问题是可以复现的。I confirm that the bug replication steps, code change instructions, and environment information have been provided, and the problem can be reproduced.
问题确认 Search before asking
Bug组件 Bug Component
Training
Bug描述 Describe the Bug
python tools/train.py -c configs/faster_rcnn/faster_rcnn_r34_vd_fpn_1x_VOC.yml --eval -o use_gpu=true
复现环境 Environment
win11 python3.7 paddlepaddle-gpu 2.3.2.post116 paddledet 2.5.0
pretrain_weights: https://paddledet.bj.bcebos.com/models/pretrained/ResNet34_vd_pretrained.pdparams
Bug描述确认 Bug description confirmation
是否愿意提交PR? Are you willing to submit a PR?