Megvii-BaseDetection / Megvii-BaseDetection/YOLOX

Does training image dimension need to be square shape for good performance

Open
#1,503 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
10.6k
Forks
2.5k
PR merge metrics
No merged PRs in 30d

Description

I am training yolox_tiny on say 320x192 image dimension and I see a 10% difference between validation score and test score. The difference is not that high when I train with a square shaped image such as 416x416 on the same dataset.

Is this expected? Have you done any testing with image input shape that are not square shape?

Contributor guide

No contributing guide indexed for this repository

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.

Research direction

Start by reproducing the YOLOX-Tiny training comparison with 320x192 and 416x416 inputs on the same dataset. Trace how non-square image dimensions are handled during training, validation, and testing, then determine whether the score difference is expected and document the finding or identify a reproducible defect.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, pytorch
Domain
computer-vision, machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.