alibaba / alibaba/EasyCV

请问如何得到自监督MoBY算法最终predictor的输出结果,直接用于以图搜图

Open
#260 1 comment 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
2k
Forks
227
PR merge metrics
No merged PRs in 30d

Description

**Describe the feature**
作者您好,当前推理时只能得到backbone的输出或者projector_q的输出,我希望可以得到最终predictor的输出,直接用于以图搜图。
在模型导出时,当前将MoBY导出为一个分类模型,舍弃了MoBY的head(即predictor),添加了一个dummy classification head。
我也可以自己实现,但是因为EasyCV包了很多层,可以实现的方式比较多,你们会官方来实现吗,或者给些建议,最elegant的实现方式。
如果我实现了可以pull request吗?
谢谢

Contributor guide

Open the contributing guide

Research direction

Start by tracing EasyCV's model export path for MoBY and compare the backbone, projector_q, predictor, and dummy classification head outputs. Confirm how the exported model currently exposes inference results, then define completion as providing the final predictor output for image retrieval without the dummy classification head. No specific files or tests are named in the issue.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, pytorch
Domain
computer-vision, machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
30/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.