File Download
There are no files associated with this item.
Supplementary
-
Citations:
- Appears in Collections:
Conference Paper: Webly Supervised Image Classification with Self-Contained Confidence
Title | Webly Supervised Image Classification with Self-Contained Confidence |
---|---|
Authors | |
Keywords | Webly supervised learning noisy labels model uncertainty |
Issue Date | 2020 |
Citation | 16th European Conference on Computer Vision (ECCV), Online, 23-28 August 2020 How to Cite? |
Abstract | This paper focuses on webly supervised learning (WSL), where datasets are built by crawling samples from the Internet and directly using search queries as web labels. Although WSL benefits from fast and low-cost data collection, noises in web labels hinder better performance of the image classification model. To alleviate this problem, in recent works, self-label supervised loss Ls is utilized together with
webly supervised loss Lw. Ls relies on pseudo labels predicted by the model itself. Since the correctness of the web label or pseudo label is usually on a case-by-case basis for each web sample, it is desirable to adjust the balance between Ls and Lw on sample level. Inspired by the ability of Deep Neural Networks (DNNs) in confidence prediction, we introduce Self-Contained Confidence (SCC) by adapting model uncertainty for WSL setting, and use it to sample-wisely balance Ls and Lw. Therefore, a simple yet effective WSL framework is proposed. A series of SCC-friendly regularization approaches are investigated, among which the proposed graph-enhanced mixup is the most effective method to provide high-quality confidence to enhance our framework. The proposed WSL framework has achieved the state-of-the-art results on two largescale WSL datasets, WebVision-1000 and Food101-N. Code is available at https://github.com/bigvideoresearch/SCC. |
Description | Poster Presentation - Paper ID: 701 ECCV 2020 take place virtually due to COVID-19 |
Persistent Identifier | http://hdl.handle.net/10722/284146 |
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Yang, JK | - |
dc.contributor.author | Feng, LT | - |
dc.contributor.author | Chen, W | - |
dc.contributor.author | Yan, X | - |
dc.contributor.author | Zheng, H | - |
dc.contributor.author | Luo, P | - |
dc.contributor.author | Zhang, W | - |
dc.date.accessioned | 2020-07-20T05:56:27Z | - |
dc.date.available | 2020-07-20T05:56:27Z | - |
dc.date.issued | 2020 | - |
dc.identifier.citation | 16th European Conference on Computer Vision (ECCV), Online, 23-28 August 2020 | - |
dc.identifier.uri | http://hdl.handle.net/10722/284146 | - |
dc.description | Poster Presentation - Paper ID: 701 | - |
dc.description | ECCV 2020 take place virtually due to COVID-19 | - |
dc.description.abstract | This paper focuses on webly supervised learning (WSL), where datasets are built by crawling samples from the Internet and directly using search queries as web labels. Although WSL benefits from fast and low-cost data collection, noises in web labels hinder better performance of the image classification model. To alleviate this problem, in recent works, self-label supervised loss Ls is utilized together with webly supervised loss Lw. Ls relies on pseudo labels predicted by the model itself. Since the correctness of the web label or pseudo label is usually on a case-by-case basis for each web sample, it is desirable to adjust the balance between Ls and Lw on sample level. Inspired by the ability of Deep Neural Networks (DNNs) in confidence prediction, we introduce Self-Contained Confidence (SCC) by adapting model uncertainty for WSL setting, and use it to sample-wisely balance Ls and Lw. Therefore, a simple yet effective WSL framework is proposed. A series of SCC-friendly regularization approaches are investigated, among which the proposed graph-enhanced mixup is the most effective method to provide high-quality confidence to enhance our framework. The proposed WSL framework has achieved the state-of-the-art results on two largescale WSL datasets, WebVision-1000 and Food101-N. Code is available at https://github.com/bigvideoresearch/SCC. | - |
dc.language | eng | - |
dc.relation.ispartof | European Conference on Computer Vision (ECCV) | - |
dc.subject | Webly supervised learning | - |
dc.subject | noisy labels | - |
dc.subject | model uncertainty | - |
dc.title | Webly Supervised Image Classification with Self-Contained Confidence | - |
dc.type | Conference_Paper | - |
dc.identifier.email | Luo, P: pluo@hku.hk | - |
dc.identifier.authority | Luo, P=rp02575 | - |
dc.identifier.hkuros | 311004 | - |