search

tensorflow / models

Models and examples built with TensorFlow
Other
76.97k stars 45.79k forks source link

how to pretrain model on MS-COCO (Deeplabv3+) #5008

Closed yswang1717 closed 2 years ago

yswang1717 commented 6 years ago

I'm training deeplabv3+ with my own backbone network and I want to pretrain my models on MS-COCO dataset. the deeplab paper shows that pre-train models on coco dataset can much improve the results, but the coco dataset just give the instance level labels, how to transfer data to satisfy the category level requirement, or anyone can give me some examples on how to pre-train models on coco datasets? and Do you have any idea how to handle overlapping segments in the MS-COCO ground truth while training the network?

Thank you for your quick reponse. The followings are details of my version. What is the top-level directory of the model you are using /home/semantic_segmentation/models/research/deeplab/ Have I written custom code No OS Platform and Distribution Ubuntu 16.04.3 LTS ( GNU/Linux 4.4.0-98-generic x86_64) TensorFlow installed from pip command TensorFlow version 1.6.0 Bazel version None CUDA/cuDNN version None GPU model and memory TITAN Xp 12GB * 8 Exact command to reproduce None.

tensorflowbutler commented 6 years ago

Thank you for your post. We noticed you have not filled out the following field in the issue template. Could you update them if they are relevant in your case, or leave them as N/A? Thanks. What is the top-level directory of the model you are using Have I written custom code OS Platform and Distribution TensorFlow installed from TensorFlow version Bazel version CUDA/cuDNN version GPU model and memory Exact command to reproduce

yswang1717 commented 6 years ago

Thank you for your quick reponse. The followings are details of my version. What is the top-level directory of the model you are using /home/semantic_segmentation/models/research/deeplab/ Have I written custom code No OS Platform and Distribution Ubuntu 16.04.3 LTS ( GNU/Linux 4.4.0-98-generic x86_64) TensorFlow installed from pip command TensorFlow version 1.6.0 Bazel version None CUDA/cuDNN version None GPU model and memory TITAN Xp 12GB * 8 Exact command to reproduce None.

VisionTheta commented 6 years ago

Maybe the paper Weakly- and Semi-Supervised Learning of a DCNN for Semantic Image Segmentation could help you.

Damanpreet commented 4 years ago

hi @yswang0522, Were you able to train Deeplab on COCO? What training configuration did you use?

kumariko commented 2 years ago

@yswang1717 We are checking to see if you still need help on this issue? Could you please have a look on the link1, link2 and let us know if it helps? Thanks!

google-ml-butler[bot] commented 2 years ago

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you.

google-ml-butler[bot] commented 2 years ago

Closing as stale. Please reopen if you'd like to work on this further.

google-ml-butler[bot] commented 2 years ago

Are you satisfied with the resolution of your issue? Yes No