prediction format

[mjanddy]

what's the Format of prediction results？

[KevinZhang1201]

you can see it at homepage of WIDER FACE and download Examples and Format of submissions.
Link: http://mmlab.ie.cuhk.edu.hk/projects/WIDERFace/support/example/Submission_example.zip

first you need to create a directory for all prediction results, like:

-- WFPred
-- 0--Parade
-- 0_Parade_marchingband_1_20.txt
-- 0_Parade_marchingband_1_74.txt
...
-- 1--Handshaking
...

In each files the detection results should be written as:
image_name
the number fo faces
x, y, w, h, confidence
...

[monkeyMirandar]

**

1. My well-trained model is a.pth file, and my well-tested model is a.pkl file. How to make the prediction format if you want to use this method for assessment?

**

[KevinZhang1201]

@monkeyMirandar Sorry, i'm not very sure what you mean. But I think you need to writte your own script to make the prediction on WIDER Face dataset with your own models and generate the prediction files as above mentioned.

[monkeyMirandar]

就是我不知道那个 prediction dir 里面应该放些什么 发自我的iPhone

…

------------------ Original ------------------ From: KevinZhang1201 <notifications@github.com> Date: Wed,Jul 31,2019 3:26 PM To: wondervictor/WiderFace-Evaluation <WiderFace-Evaluation@noreply.github.com> Cc: monkeyMirandar <1391069856@qq.com>, Mention <mention@noreply.github.com> Subject: Re: [wondervictor/WiderFace-Evaluation] prediction format (#3) @monkeyMirandar Sorry, i'm not very sure what you mean. But I think you need to writte your own script to make the prediction on WIDER Face dataset with your own models and generate the prediction files as above mentioned. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

[KevinZhang1201]

Prediction dir
└─── 0--Parade
├───└─── 0_Parade_marchingband_1_20.txt
├───├─── 0_Parade_marchingband_1_74.txt
├───├─── ...
├─── 1--Handshaking
├───└─── 1_Handshaking_Handshaking_1_35.txt
├───├─── 1_Handshaking_Handshaking_1_94.txt
├───├─── ...
├─── ...
the files should be organized as this.

[monkeyMirandar]

我的训练模型已经出来了！怎么才可以得到这个文件夹呀？可以具体说说吗？ 发自我的iPhone

…

------------------ Original ------------------ From: KevinZhang1201 <notifications@github.com> Date: Wed,Jul 31,2019 5:17 PM To: wondervictor/WiderFace-Evaluation <WiderFace-Evaluation@noreply.github.com> Cc: monkeyMirandar <1391069856@qq.com>, Mention <mention@noreply.github.com> Subject: Re: [wondervictor/WiderFace-Evaluation] prediction format (#3) Prediction dir └─── 0--Parade ├───└─── 0_Parade_marchingband_1_20.txt ├───├─── 0_Parade_marchingband_1_74.txt ├───├─── ... ├─── 1--Handshaking ├───└─── 1_Handshaking_Handshaking_1_35.txt ├───├─── 1_Handshaking_Handshaking_1_94.txt ├───├─── ... ├─── ... the files should be organized as this. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

[KevinZhang1201]

if you use python to test your model, you can write it like this:

1. iterate all images in wider face dataset
2. apply your to each images, generate the prediction results and write them into cooresponding file

[foocker]

我的训练模型已经出来了！怎么才可以得到这个文件夹呀？可以具体说说吗？ 发自我的iPhone
…
------------------ Original ------------------ From: KevinZhang1201 notifications@github.com Date: Wed,Jul 31,2019 5:17 PM To: wondervictor/WiderFace-Evaluation WiderFace-Evaluation@noreply.github.com Cc: monkeyMirandar 1391069856@qq.com, Mention mention@noreply.github.com Subject: Re: [wondervictor/WiderFace-Evaluation] prediction format (#3) Prediction dir └─── 0--Parade ├───└─── 0_Parade_marchingband_1_20.txt ├───├─── 0_Parade_marchingband_1_74.txt ├───├─── ... ├─── 1--Handshaking ├───└─── 1_Handshaking_Handshaking_1_35.txt ├───├─── 1_Handshaking_Handshaking_1_94.txt ├───├─── ... ├─── ... the files should be organized as this. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

....

[foocker]

if you use python to test your model, you can write it like this:

1. iterate all images in wider face dataset

2. apply your to each images, generate the prediction results and write them into cooresponding file

the *.mat file are corresponding test dataset? why not val dataset. so, evaluate on val dataset, you should rewrite evaluation code?

i have test on val dataset, it's work! but i'm not plan to read the evaluate.py . so, what's logic of evaluate.py? such ,
easy, midum, hard, in *.mat, but val dataset has no these class, how it evaluate results that preddicted from val dataset, for *.mat easy,midum, hard class, but in *.mat.

[Hiteshsaai]

can the prediction results be in text format?, meanwhile the ground truth is in .mat format

[wondervictor]

prediction results are the same as the submission examples (txt files)

[NguyenKhacTuanAnh]

prediction results are the same as the submission examples (txt files)

why do it have this error? I have run python setup.py build_ext --inplace
and python3 evaluation.py -p results/ -g ground_truth/

Reading Predictions : 100%|█████████████████████████████████████████████████████| 61/61 [00:00<00:00, 603.65it/s]
Processing easy: 100%|█████████████████████████████████████████████████████████| 61/61 [00:00<00:00, 4502.23it/s]
Processing medium: 100%|███████████████████████████████████████████████████████| 61/61 [00:00<00:00, 4714.26it/s]
Processing hard: 100%|█████████████████████████████████████████████████████████| 61/61 [00:00<00:00, 4092.99it/s]
==================== Results ====================
Easy Val AP: 0.0
Medium Val AP: 0.0
Hard Val AP: 0.0