Generalized category discovery
<p>In this paper, we consider a highly general image recognition setting wherein, given a labelled and unlabelled set of images, the task is to categorize all images in the unlabelled set. Here, the unlabelled images may come from labelled classes or from novel ones. Existing recognition metho...
| Main Authors: | , , , |
|---|---|
| Format: | Conference item |
| Language: | English |
| Published: |
IEEE
2022
|
| _version_ | 1797107911342686208 |
|---|---|
| author | Vaze, S Han, K Vedaldi, A Zisserman, A |
| author_facet | Vaze, S Han, K Vedaldi, A Zisserman, A |
| author_sort | Vaze, S |
| collection | OXFORD |
| description | <p>In this paper, we consider a highly general image recognition setting wherein, given a labelled and unlabelled set of images, the task is to categorize all images in the unlabelled set. Here, the unlabelled images may come from labelled classes or from novel ones. Existing recognition methods are not able to deal with this setting, because they make several restrictive assumptions, such as the unlabelled instances only coming from known – or unknown – classes, and the number of unknown classes being known a-priori. We address the more unconstrained setting, naming it ‘Generalized Category Discovery’, and challenge all these assumptions. We first establish strong baselines by taking state-of-the-art algorithms from novel category discovery and adapting them for this task. Next, we propose the use of vision transformers with contrastive representation learning for this open-world setting. We then introduce a simple yet effective semi-supervised k-means method to cluster the unlabelled data into seen and unseen classes automatically, substantially outperforming the baselines. Finally, we also propose a new approach to estimate the number of classes in the unlabelled data. We thoroughly evaluate our approach on public datasets for generic object classification and on fine-grained datasets, leveraging the recent Semantic Shift Benchmark suite. Code: https://www.robots.ox.ac.uk/~vgg/research/gcd</p> |
| first_indexed | 2024-03-07T07:22:21Z |
| format | Conference item |
| id | oxford-uuid:15f42099-2a22-4808-8797-b6256e1b5bf0 |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-03-07T07:22:21Z |
| publishDate | 2022 |
| publisher | IEEE |
| record_format | dspace |
| spelling | oxford-uuid:15f42099-2a22-4808-8797-b6256e1b5bf02022-10-26T09:53:55ZGeneralized category discoveryConference itemhttp://purl.org/coar/resource_type/c_5794uuid:15f42099-2a22-4808-8797-b6256e1b5bf0EnglishSymplectic ElementsIEEE2022Vaze, SHan, KVedaldi, AZisserman, A<p>In this paper, we consider a highly general image recognition setting wherein, given a labelled and unlabelled set of images, the task is to categorize all images in the unlabelled set. Here, the unlabelled images may come from labelled classes or from novel ones. Existing recognition methods are not able to deal with this setting, because they make several restrictive assumptions, such as the unlabelled instances only coming from known – or unknown – classes, and the number of unknown classes being known a-priori. We address the more unconstrained setting, naming it ‘Generalized Category Discovery’, and challenge all these assumptions. We first establish strong baselines by taking state-of-the-art algorithms from novel category discovery and adapting them for this task. Next, we propose the use of vision transformers with contrastive representation learning for this open-world setting. We then introduce a simple yet effective semi-supervised k-means method to cluster the unlabelled data into seen and unseen classes automatically, substantially outperforming the baselines. Finally, we also propose a new approach to estimate the number of classes in the unlabelled data. We thoroughly evaluate our approach on public datasets for generic object classification and on fine-grained datasets, leveraging the recent Semantic Shift Benchmark suite. Code: https://www.robots.ox.ac.uk/~vgg/research/gcd</p> |
| spellingShingle | Vaze, S Han, K Vedaldi, A Zisserman, A Generalized category discovery |
| title | Generalized category discovery |
| title_full | Generalized category discovery |
| title_fullStr | Generalized category discovery |
| title_full_unstemmed | Generalized category discovery |
| title_short | Generalized category discovery |
| title_sort | generalized category discovery |
| work_keys_str_mv | AT vazes generalizedcategorydiscovery AT hank generalizedcategorydiscovery AT vedaldia generalizedcategorydiscovery AT zissermana generalizedcategorydiscovery |