同步操作将从 PaddlePaddle/PaddleSeg 强制同步,此操作会覆盖自 Fork 仓库以来所做的任何修改,且无法恢复!!!
确定后同步将在后台操作,完成时将刷新页面,请耐心等待。
Official resource for the paper PP-HumanSeg: Connectivity-Aware Portrait Segmentation With a Large-Scale Teleconferencing Video Dataset. [Paper | Poster | YouTube]
SCL (Semantic Connectivity-aware Learning) framework, which introduces a SC Loss (Semantic Connectivity-aware Loss) to improve the quality of segmentation results from the perspective of connectivity. SCL can improve the integrity of segmentation objects and increase segmentation accuracy. Support multi-class segmentation. [Source code]
(a) It indicates prediction and ground truth, i.e. P and G. (b) Connected components are generated through the CCL algorithm, respectively. (c) Connected components are matched using the IoU value.
The experimental results on our Teleconferencing Video Dataset are shown in paper, and the experimental results on Cityscapes are as follows:
Model | Backbone | Learning Strategy | GPUs * Batch Size(Per Card) | Training Iters | mIoU (%) | Config |
---|---|---|---|---|---|---|
OCRNet | HRNet-W48 | - | 2*2 | 40000 | 76.23 | config |
OCRNet | HRNet-W48 | SCL | 2*2 | 40000 | 78.29(+2.06) | config |
FCN | HRNet-W18 | - | 2*4 | 80000 | 77.81 | config |
FCN | HRNet-W18 | SCL | 2*4 | 80000 | 78.68(+0.87) | config |
Fast SCNN | - | - | 2*4 | 40000 | 56.41 | config |
Fast SCNN | - | SCL | 2*4 | 40000 | 57.37(+0.96) | config |
A large-scale video portrait dataset that contains 291 videos from 23 conference scenes with 14K frames. This dataset contains various teleconferencing scenes, various actions of the participants, interference of passers-by and illumination change. The data can be obtained by sending an email to paddleseg@baidu.com via an official email (not use qq, gmail, etc.) including your institution/company information and the purpose on the dataset.
If our project is useful in your research, please citing:
@InProceedings{Chu_2022_WACV,
author = {Chu, Lutao and Liu, Yi and Wu, Zewu and Tang, Shiyu and Chen, Guowei and Hao, Yuying and Peng, Juncai and Yu, Zhiliang and Chen, Zeyu and Lai, Baohua and Xiong, Haoyi},
title = {PP-HumanSeg: Connectivity-Aware Portrait Segmentation With a Large-Scale Teleconferencing Video Dataset},
booktitle = {Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) Workshops},
month = {January},
year = {2022},
pages = {202-209}
}
此处可能存在不合适展示的内容,页面不予展示。您可通过相关编辑功能自查并修改。
如您确认内容无涉及 不当用语 / 纯广告导流 / 暴力 / 低俗色情 / 侵权 / 盗版 / 虚假 / 无价值内容或违法国家有关法律法规的内容,可点击提交进行申诉,我们将尽快为您处理。