Zero-Shot Scene Change Detection
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910790954516480 |
|---|---|
| author | Cho, Kyusik Kim, Dong Yeop Kim, Euntai |
| author_facet | Cho, Kyusik Kim, Dong Yeop Kim, Euntai |
| contents | We present a novel, training-free approach to scene change detection. Our method leverages tracking models, which inherently perform change detection between consecutive frames of video by identifying common objects and detecting new or missing objects. Specifically, our method takes advantage of the change detection effect of the tracking model by inputting reference and query images instead of consecutive frames. Furthermore, we focus on the content gap and style gap between two input images in change detection, and address both issues by proposing adaptive content threshold and style bridging layers, respectively. Finally, we extend our approach to video, leveraging rich temporal information to enhance the performance of scene change detection. We compare our approach and baseline through various experiments. While existing train-based baseline tend to specialize only in the trained domain, our method shows consistent performance across various domains, proving the competitiveness of our approach. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_11210 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Zero-Shot Scene Change Detection Cho, Kyusik Kim, Dong Yeop Kim, Euntai Computer Vision and Pattern Recognition We present a novel, training-free approach to scene change detection. Our method leverages tracking models, which inherently perform change detection between consecutive frames of video by identifying common objects and detecting new or missing objects. Specifically, our method takes advantage of the change detection effect of the tracking model by inputting reference and query images instead of consecutive frames. Furthermore, we focus on the content gap and style gap between two input images in change detection, and address both issues by proposing adaptive content threshold and style bridging layers, respectively. Finally, we extend our approach to video, leveraging rich temporal information to enhance the performance of scene change detection. We compare our approach and baseline through various experiments. While existing train-based baseline tend to specialize only in the trained domain, our method shows consistent performance across various domains, proving the competitiveness of our approach. |
| title | Zero-Shot Scene Change Detection |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2406.11210 |