tensorflow / tensorflow/models

Evaluation accuracy with frozen graph lower than value given Deeplabv3+ Xception 65 baackbone

Open
#10,242 0 comments 0 reactions 2 assignees View on GitHub

@aquariusjay is already working on this.

Since Sep 7, 2021.

models:research type:support
Dominant language
Python
Stars
77.7k
Forks
44.8k
PR merge metrics
No merged PRs in 30d

Description

When using the xception65_cityscapes_trainfine frozen graph alone instead of using the entire checkpoint provided to calculate the validation accuracy on the Cityscapes dataset with Eval scales of 1, I receive a total mIOU score of 63%, which is quite a bit lower than the stated mIOU score of 78.79%. I have used the same way of preprocessing and calculating accuracy as the eval.py script. Is this difference in accuracy known, and is there a difference between the frozen graph given and the model that scored 78.79%.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.