Gespeichert in:
| Hauptverfasser: | , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | https://arxiv.org/abs/2410.05935 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866912064314802176 |
|---|---|
| author | Taniguchi, Takara Furuta, Ryosuke |
| author_facet | Taniguchi, Takara Furuta, Ryosuke |
| contents | We tackle one-shot object detection in Japanese Manga. The rising global popularity of Japanese manga has made the object detection of character faces increasingly important, with potential applications such as automatic colorization. However, obtaining sufficient data for training conventional object detectors is challenging due to copyright restrictions. Additionally, new characters appear every time a new volume of manga is released, making it impractical to re-train object detectors each time to detect these new characters. Therefore, one-shot object detection, where only a single query (reference) image is required to detect a new character, is an essential task in the manga industry. One challenge with one-shot object detection in manga is the large variation in the poses and facial expressions of characters in target images, despite having only one query image as a reference. Another challenge is that the frequency of character appearances follows a long-tail distribution. To overcome these challenges, we propose a data augmentation method in feature space to increase the variation of the query. The proposed method augments the feature from the query by adding Gaussian noise, with the noise variance at each channel learned during training. The experimental results show that the proposed method improves the performance for both seen and unseen classes, surpassing data augmentation methods in image space. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_05935 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Learning Gaussian Data Augmentation in Feature Space for One-shot Object Detection in Manga Taniguchi, Takara Furuta, Ryosuke Computer Vision and Pattern Recognition Multimedia We tackle one-shot object detection in Japanese Manga. The rising global popularity of Japanese manga has made the object detection of character faces increasingly important, with potential applications such as automatic colorization. However, obtaining sufficient data for training conventional object detectors is challenging due to copyright restrictions. Additionally, new characters appear every time a new volume of manga is released, making it impractical to re-train object detectors each time to detect these new characters. Therefore, one-shot object detection, where only a single query (reference) image is required to detect a new character, is an essential task in the manga industry. One challenge with one-shot object detection in manga is the large variation in the poses and facial expressions of characters in target images, despite having only one query image as a reference. Another challenge is that the frequency of character appearances follows a long-tail distribution. To overcome these challenges, we propose a data augmentation method in feature space to increase the variation of the query. The proposed method augments the feature from the query by adding Gaussian noise, with the noise variance at each channel learned during training. The experimental results show that the proposed method improves the performance for both seen and unseen classes, surpassing data augmentation methods in image space. |
| title | Learning Gaussian Data Augmentation in Feature Space for One-shot Object Detection in Manga |
| topic | Computer Vision and Pattern Recognition Multimedia |
| url | https://arxiv.org/abs/2410.05935 |