Custom Bisenetv2 config for new dataset

[JJLimmm]

Thanks for your contribution and we appreciate it a lot. The following instructions would make your pull request more healthy and more easily get feedback. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers.

Motivation

New project that i am diving into, since AV is the future and testing out a semantic segmentation model to detect water puddles on the road while driving.

Modification

Added in a Custom dataset class called waterpuddlesDataset() with 2 classes; background and waterpuddles.

1. Added in waterpuddles.py into configs/base/dataset/, the new dataset is following the ChaseDB1 format. Users can convert the dataset from the original public Puddle-1000 dataset and use the ChaseDB1.py under convert_datasets/ to get the required format aligned with ChaseDB1.
2. Added custom dataset class into mmseg/datasets/ directory called waterpuddles.py
3. Updated class_names.py in mmseg/core/evaluation/ directory to include the waterpuddles dataset attributes.

BC-breaking (Optional)

Backward-Compatible

Checklist

1. Pre-commit or other linting tools are used to fix the potential lint issues.
2. The modification is covered by complete unit tests. If not, please add more unit test to ensure the correctness.
3. If the modification has potential influence on downstream projects, this PR should be tested with downstream projects, like MMDet or MMDet3D.
4. The documentation has been modified accordingly, like docstring or example tutorials.

[CLAassistant]

All committers have signed the CLA.

[MeowZheng]

Many thanks for your contribution, but I strongly recommend that you might split this PR into 2 PRs, one is to add the 'warterpuddles' dataset and its cfg file, and the other is for adding new training cfg for bisenetv2.

BTW, please sign CLA

[MengzhangLI]

Hi, @JJLimmm Thanks so much for your nice PR.

It is better to support this feature by two steps:

First, support water puddle dataset. Please refer to support Vaihingen dataset:#1171.

Second, add BiSeNetV2 for this dataset for a new benchmark, please refer to add FastFCN on ADE20K dataset. #972.

In my point of view, this PR would better just support waterpuddle dataset. Because I think many parts of your config files need to be modified to satisfy merging master branch.

But its OK. Feel free to contact us if you meet any problems.

Best,

[MengzhangLI]

Too many comment paragraphs. Just keep L4-L9.

[codecov]

Codecov Report

Merging #1425 (017bb45) into master (30864ea) will decrease coverage by 0.11%.
The diff coverage is 15.38%.

❗ Current head 017bb45 differs from pull request most recent head 69560f6. Consider uploading reports for the commit 69560f6 to get more accurate results

@@            Coverage Diff             @@
##           master    #1425      +/-   ##
==========================================
- Coverage   90.22%   90.10%   -0.12%     
==========================================
  Files         136      137       +1     
  Lines        7946     7956      +10     
  Branches     1323     1324       +1     
==========================================
  Hits         7169     7169              
- Misses        552      562      +10     
  Partials      225      225              

Flag	Coverage Δ
unittests	90.10% <15.38%> (-0.12%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
mmseg/datasets/waterpuddles.py	0.00% <0.00%> (ø)
mmseg/core/evaluation/class_names.py	88.33% <66.66%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update d6a6c39...69560f6. Read the comment docs.

[OpenMMLab-Assistant003]

Hi @JJLimmm！We are grateful for your efforts in helping improve this open-source project during your personal time.
Welcome to join OpenMMLab Special Interest Group (SIG) private channel on Discord, where you can share your experiences, ideas, and build connections with like-minded peers. To join the SIG channel, simply message moderator— OpenMMLab on Discord or briefly share your open-source contributions in the #introductions channel and we will assist you. Look forward to seeing you there! Join us ：https://discord.gg/UjgXkPWNqA
If you have a WeChat account，welcome to join our community on WeChat. You can add our assistant ：openmmlabwx. Please add "mmsig + Github ID" as a remark when adding friends：）
Thank you again for your contribution❤ @JJLimmm