Bringing the Context Back into Object Recognition, Robustly
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866917950761467904 |
|---|---|
| author | Janouskova, Klara Gavrus, Cristian Matas, Jiri |
| author_facet | Janouskova, Klara Gavrus, Cristian Matas, Jiri |
| contents | In object recognition, both the subject of interest (referred to as foreground, FG, for simplicity) and its surrounding context (background, BG) may play an important role. However, standard supervised learning often leads to unintended over-reliance on the BG, limiting model robustness in real-world deployment settings. The problem is mainly addressed by suppressing the BG, sacrificing context information for improved generalization.
We propose "Localize to Recognize Robustly" (L2R2), a novel recognition approach which exploits the benefits of context-aware classification while maintaining robustness to distribution shifts. L2R2 leverages advances in zero-shot detection to localize the FG before recognition. It improves the performance of both standard recognition with supervised training, as well as multimodal zero-shot recognition with VLMs, while being robust to long-tail BGs and distribution shifts. The results confirm localization before recognition is possible for a wide range of datasets and they highlight the limits of object detection on others |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2411_15933 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Bringing the Context Back into Object Recognition, Robustly Janouskova, Klara Gavrus, Cristian Matas, Jiri Computer Vision and Pattern Recognition In object recognition, both the subject of interest (referred to as foreground, FG, for simplicity) and its surrounding context (background, BG) may play an important role. However, standard supervised learning often leads to unintended over-reliance on the BG, limiting model robustness in real-world deployment settings. The problem is mainly addressed by suppressing the BG, sacrificing context information for improved generalization. We propose "Localize to Recognize Robustly" (L2R2), a novel recognition approach which exploits the benefits of context-aware classification while maintaining robustness to distribution shifts. L2R2 leverages advances in zero-shot detection to localize the FG before recognition. It improves the performance of both standard recognition with supervised training, as well as multimodal zero-shot recognition with VLMs, while being robust to long-tail BGs and distribution shifts. The results confirm localization before recognition is possible for a wide range of datasets and they highlight the limits of object detection on others |
| title | Bringing the Context Back into Object Recognition, Robustly |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2411.15933 |