onnx / onnx/models

Improve accuracy of the ultraface-rfb-640.onnx model

Open
#630 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

question
Dominant language
Jupyter Notebook
Stars
9.8k
Forks
1.6k
PR merge metrics
No merged PRs in 30d

Description

Ask a Question

Question

Hi,

Currently, I'm using version-RFB-640.onnx to perform face detection on a video from a cpp program. It works well on some videos but on others, it doesn't. The detection seems to be intermittent, at some frames, it detects some of the faces, but most of the time it doesn't. Is there a way to maybe adjust the tolerance/confidence of the model or improve the accuracy of the model?

Frame where onnx model did detect the face (the girl with purple blazer):

accurate_onnx

Frame where it didn't detect the face:

not_accurate_onnx

As you can see, it doesn't seem to be working on the more further faces or from the guys on the right.

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 intermittent detections with version-RFB-640.onnx in the reported C++ program, comparing the supplied successful and failing frames. Check whether inference settings or image characteristics explain the missed faces; done would require a clearly identified accuracy improvement or documented limitation, but the issue names no repository files or tests.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
computer-vision, machine-learning
Issue type
Bug
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.