All person ReIDentification samples. by X-ing · Pull Request #19108 · opencv/opencv

X-ing · 2020-12-14T13:15:08Z

Hi, this is a sample of person ReIDentification.
This sample uses one of the state-of-the-art ReID models built by Tencent YoutuLab.

Status Update

Currently there are a python script and a Cpp script with the same features, both supportting extracting ReID feature with youtu baseline ReID model.

Introduction

This sample takes two sets of images as input, namely the query set and gallery set.

This sample first load the onnx model of youtu ReID baseline and then extract the ReID feature of all images in the query and gallery set.

For each query image, the gallery images are ranked based on its consine similarity with the query image.

Finally the ranking list is visualized in an output folder.
Following is the example of the visualization results. The first image of each row is the query, and the rest is the top-10 most similar gallery image found by the ReID model.

Reference

Zheng F, Deng C, Sun X, et al. Pyramidal person re-identification via multi-loss dynamic training[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2019: 8514-8522.

Yu F, Jiang X, Gong Y, et al. Devil's in the Details: Aligning Visual Clues for Conditional Embedding in Person Re-Identification[J]. arXiv e-prints, 2020: arXiv: 2009.05250.

Pull Request Readiness Checklist

See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request

I agree to contribute to the project under Apache 2 License.
To the best of my knowledge, the proposed patch is not based on a code under GPL or other license that is incompatible with OpenCV
The PR is proposed to proper branch
There is reference to original bug report and related work
There is accuracy test, performance test and test data in opencv_extra repository, if applicable
Patch to opencv_extra has the same branch name.
The feature is well documented and sample code can be built with the project CMake

force_builders_only=linux,windows,mac,docs
**WIP**
test_modules=none

dkurt · 2020-12-14T13:51:42Z

samples/dnn/person_reid.cpp

+		features.push_back(normalization(temp_feature));
+	    }
+	}
+	return features;


please fix formatting

alalek

This sample is really hard to use due to missing of input data

alalek · 2020-12-14T15:33:16Z

samples/dnn/person_reid.cpp

+// 
+// 
+//                           License Agreement
+//                 For Open Source Computer Vision Library


OpenCV samples don't use license headers.
These files are used as a part of documentation which may use whole sample's file or its snippets.

You may left here short notice about sample's authors and/or model's authors (and/or model's limitations).

https://github.com/opencv/opencv/wiki/Coding_Style_Guide#file-structure

alalek · 2020-12-14T15:33:49Z

samples/dnn/person_reid.cpp

+"{backend b  | 0               | Choose one of computation backends: "
+"0: automatically (by default), "
+"1: Halide language (http://halide-lang.org/), "
+"2: Intel's Deep Learning Inference Engine (https://software.intel.com/openvino-toolkit), "
+"3: OpenCV implementation }"
+"{target  t  | 0                | Choose one of target computation devices: "


CUDA backend?

samples/dnn/person_reid.cpp

alalek · 2020-12-14T15:34:26Z

samples/dnn/person_reid.cpp

+        {
+            sum += feature[i] * feature[i];
+	}
+	sum = sqrt(sum);
+	for(int i = 0; i < (int)feature.size(); i++)


fix indentation

samples/dnn/person_reid.cpp

alalek · 2020-12-14T15:36:03Z

samples/dnn/person_reid.cpp

+	return ret;
+    }
+
+    static Mat drawRankList(std::string queryName, std::vector<std::string> galleryImageNames, std::vector<int> topk_index, const int resize_h, const int resize_w)


std::string
std::vector<std::string>

use references, const references instead of copied values.

alalek · 2020-12-14T15:40:42Z

samples/dnn/person_reid.cpp

+"{resize_h   | 256             | height of input image}"
+"{resize_w   | 128             | width of input image}"


of input image

IMHO, it doesn't look accurate.
This is about size of the input for model inference.

hello, i have fixed all indentation mistakes, used references for parameters, and changed the description of resize_* to size of model.

X-ing · 2020-12-16T03:24:44Z

This sample is really hard to use due to missing of input data

we have opened a new repo for model and sample input images.

sl-sergei

Thank you! 👍

alalek · 2021-01-12T18:27:05Z

root and others added 10 commits
root
root <root@TENCENT64.site>

Please cleanup duplicated commits and their malformed emails before PR merge.
"Merge" commits should not be used in the Pull Requests. Use git rebase instead.

X-ing · 2021-01-13T03:35:37Z

root and others added 10 commits
root
root <root@TENCENT64.site>

Please cleanup duplicated commits and their malformed emails before PR merge.
"Merge" commits should not be used in the Pull Requests. Use git rebase instead.

We have squashed all the commits into one.

dkurt reviewed Dec 14, 2020

View reviewed changes

alalek reviewed Dec 14, 2020

View reviewed changes

xinyangj force-pushed the sample_person_reid branch from 7e1fd46 to baa635f Compare December 15, 2020 04:34

X-ing requested review from alalek and dkurt December 16, 2020 11:27

asmorkalov requested a review from sl-sergei December 25, 2020 09:52

sl-sergei approved these changes Jan 11, 2021

View reviewed changes

sample of person ReIDentification.

f9f33a7

mznw force-pushed the sample_person_reid branch from 04e336b to f9f33a7 Compare January 13, 2021 03:26

X-ing closed this Jan 13, 2021

X-ing reopened this Jan 13, 2021

alalek assigned sl-sergei Jan 13, 2021

opencv-pushbot merged commit bbdae29 into opencv:master Jan 13, 2021

alalek mentioned this pull request Apr 9, 2021

(5.x) Merge 4.x #19885

Merged

		"{resize_h \| 256 \| height of input image}"
		"{resize_w \| 128 \| width of input image}"

Uh oh!

Conversation

X-ing commented Dec 14, 2020 • edited by alalek Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Status Update

Introduction

Reference

Pull Request Readiness Checklist

Uh oh!

dkurt Dec 14, 2020

Choose a reason for hiding this comment

Uh oh!

alalek left a comment

Choose a reason for hiding this comment

Uh oh!

alalek Dec 14, 2020

Choose a reason for hiding this comment

Uh oh!

alalek Dec 14, 2020

Choose a reason for hiding this comment

Uh oh!

Uh oh!

alalek Dec 14, 2020

Choose a reason for hiding this comment

Uh oh!

Uh oh!

alalek Dec 14, 2020

Choose a reason for hiding this comment

Uh oh!

alalek Dec 14, 2020

Choose a reason for hiding this comment

Uh oh!

mznw Dec 17, 2020

Choose a reason for hiding this comment

Uh oh!

X-ing commented Dec 16, 2020 • edited by alalek Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sl-sergei left a comment

Choose a reason for hiding this comment

Uh oh!

alalek commented Jan 12, 2021

Uh oh!

X-ing commented Jan 13, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

X-ing commented Dec 14, 2020 •

edited by alalek

Loading

X-ing commented Dec 16, 2020 •

edited by alalek

Loading

X-ing commented Jan 13, 2021 •

edited

Loading