ValueError while fine-tuning TensorFlow Object Detection model

Question

I'm training an object detection model on a custom dataset and fine-tuning ssd_inception_v2 pre-trained model from TensorFlow model zoo for the same. The training script is pointed to the model.ckpt.index given in the tarball.

TensorFlow returns a ValueError exception (No variables to save) with warnings about weights not being present:

WARNING:root:Variable [FeatureExtractor/InceptionV2/Mixed_5c_2_Conv2d_5_3x3_s2_128/BatchNorm/beta/RMSProp_1] is not available in checkpoint                                                                  [0/1875]
WARNING:root:Variable [FeatureExtractor/InceptionV2/Mixed_5c_2_Conv2d_5_3x3_s2_128/BatchNorm/gamma] is not available in checkpoint
WARNING:root:Variable [FeatureExtractor/InceptionV2/Mixed_5c_2_Conv2d_5_3x3_s2_128/BatchNorm/gamma/ExponentialMovingAverage] is not available in checkpoint
WARNING:root:Variable [FeatureExtractor/InceptionV2/Mixed_5c_2_Conv2d_5_3x3_s2_128/BatchNorm/gamma/RMSProp] is not available in checkpoint
WARNING:root:Variable [FeatureExtractor/InceptionV2/Mixed_5c_2_Conv2d_5_3x3_s2_128/BatchNorm/gamma/RMSProp_1] is not available in checkpoint
WARNING:root:Variable [FeatureExtractor/InceptionV2/Mixed_5c_2_Conv2d_5_3x3_s2_128/BatchNorm/moving_mean] is not available in checkpoint
WARNING:root:Variable [FeatureExtractor/InceptionV2/Mixed_5c_2_Conv2d_5_3x3_s2_128/BatchNorm/moving_variance] is not available in checkpoint
WARNING:root:Variable [FeatureExtractor/InceptionV2/Mixed_5c_2_Conv2d_5_3x3_s2_128/weights] is not available in checkpoint
WARNING:root:Variable [FeatureExtractor/InceptionV2/Mixed_5c_2_Conv2d_5_3x3_s2_128/weights/ExponentialMovingAverage] is not available in checkpoint
WARNING:root:Variable [FeatureExtractor/InceptionV2/Mixed_5c_2_Conv2d_5_3x3_s2_128/weights/RMSProp] is not available in checkpoint
WARNING:root:Variable [FeatureExtractor/InceptionV2/Mixed_5c_2_Conv2d_5_3x3_s2_128/weights/RMSProp_1] is not available in checkpoint

and more such errors. I have also tried ssd_mobilenet_v2 with the same errors. I have just changed the number of classes in the config file along with the "PATHS_TO_BE_CONFIGURED".

Using TensorFlow 1.7 on GPU.

I was able to resolve this by pointing to model.ckpt instead of model.ckpt.index Yes, the file model.ckpt doesn't exist but you need to point it that way so that all 3 files are picked. Strange this wasn't documented anywhere. — Anupam Sobti
– Anupam Sobti, Commented May 3, 2018 at 9:38

Jason Lau · Accepted Answer · 2018-05-03 07:14:58Z

1

Got the same problem on windows 7 with TensorFlow 1.7 on CPU. If 'fine tune' is off, the codes run without errors.

answered May 3, 2018 at 7:14

Jason Lau

111 bronze badge

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

ValueError while fine-tuning TensorFlow Object Detection model

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related