我正在使用ssd_mobilenet_v1_coco.config和
经过计划培训后,我将num_classes的值改为20,增加了13件东西。
python model_main.py --alsologtostderr --model_dir=training/ --pipeline_config_path=training/ssd_mobilenet_v1_coco.config我一直试着用命令学习,但我遇到了一个错误。要增加num_classes,我应该做什么?我应该从一开始就抓起num_classes=100吗?我需要帮助。
model {
ssd {
num_classes: 20
box_coder {
faster_rcnn_box_coder {
y_scale: 10.0
x_scale: 10.0
height_scale: 5.0
width_scale: 5.0
}
File "/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/training/saver.py", line 1326, in restore
err, "a mismatch between the current graph and the graph")
tensorflow.python.framework.errors_impl.InvalidArgumentError: Restoring from checkpoint failed. This is most likely due to a mismatch between the current graph and the graph from the checkpoint. Please ensure that you have not altered the graph expected based on the checkpoint. Original error:
Assign requires shapes of both tensors to match. lhs shape= [126] rhs shape= [84]
[[node save/Assign_56 (defined at /usr/local/lib/python3.6/dist-packages/tensorflow_core/python/framework/ops.py:1748) ]]发布于 2020-09-16 13:49:51
我最近也遇到了类似的问题。为了解决我的问题,我必须做以下工作:
python research/object_detection/model_main.py \
--model_dir=./model/finetune0 \
--pipeline_config_path=./model/pipeline.config \
--alsologtostderr我的档案结构:
+ models
-+ model
--+ checkpoint
--+ model.ckpt.index
--+ model.ckpt.meta
--+ model.ckpt.data-00000-of-00001
--+ pipeline.config
--- finetune0 (will be autogenerated)
-- data (tfrecord dataset)
-- annotations (labels)
...上下文
看起来,当您已经在model_dir上有一个检查点时,脚本将尝试恢复对所提供的模型的培训,但是pipeline.config上的新配置将与当前模型不匹配(num_class不同)。
如果您在fine_tune_checkpoint中提供了这个检查点并将model_dir指向一个新文件夹,它将从检查点变量构建模型,调整它以匹配新的配置,然后开始培训。
https://stackoverflow.com/questions/63137975
复制相似问题