search

tanakataiki / ssd_kerasV2

BetterNetWork4SSD
MIT License
144 stars 56 forks source link

Wrong Layer Choosen for SSD from MobilenetV2 backbone #27

Open prateethvnayak opened 5 years ago

prateethvnayak commented 5 years ago

The line here corresponds to the 12th block output.

The correct replacement for it is FeatureExtractor = Model(inputs=mobilenetv2.input, outputs=mobilenetv2.get_layer('block_12_add').output)

wyjeong commented 4 years ago

@prateethvnayak Thanks, your modification works. By the way, didn't you face this problem when you train your model?

InvalidArgumentError: 2 root error(s) found.
  (0) Invalid argument: Incompatible shapes: [16,7308,4] vs. [16,1692,4]
     [[{{node loss/predictions_loss/sub}}]]
     [[loss/mul/_3917]]
  (1) Invalid argument: Incompatible shapes: [16,7308,4] vs. [16,1692,4]
     [[{{node loss/predictions_loss/sub}}]]
0 successful operations.
0 derived errors ignored.
prateethvnayak commented 4 years ago

@wyjeong No, did you make some changes ? it seems to be an error in the loss module. I cannot comment more without looking at your code