Shifted Autoencoders for Point Annotation Restoration in Object Counting
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2023
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866916159404638208 |
|---|---|
| author | Zou, Yuda Xiao, Xin Zhou, Peilin Sun, Zhichao Du, Bo Xu, Yongchao |
| author_facet | Zou, Yuda Xiao, Xin Zhou, Peilin Sun, Zhichao Du, Bo Xu, Yongchao |
| contents | Object counting typically uses 2D point annotations. The complexity of object shapes and the subjectivity of annotators may lead to annotation inconsistency, potentially confusing counting model training. Some sophisticated noise-resistance counting methods have been proposed to alleviate this issue. Differently, we aim to directly refine the initial point annotations before training counting models. For that, we propose the Shifted Autoencoders (SAE), which enhances annotation consistency. Specifically, SAE applies random shifts to initial point annotations and employs a UNet to restore them to their original positions. Similar to MAE reconstruction, the trained SAE captures general position knowledge and ignores specific manual offset noise. This allows to restore the initial point annotations to more general and thus consistent positions. Extensive experiments show that using such refined consistent annotations to train some advanced (including noise-resistance) object counting models steadily/significantly boosts their performances. Remarkably, the proposed SAE helps to set new records on nine datasets. We will make codes and refined point annotations available. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2312_07190 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Shifted Autoencoders for Point Annotation Restoration in Object Counting Zou, Yuda Xiao, Xin Zhou, Peilin Sun, Zhichao Du, Bo Xu, Yongchao Computer Vision and Pattern Recognition Object counting typically uses 2D point annotations. The complexity of object shapes and the subjectivity of annotators may lead to annotation inconsistency, potentially confusing counting model training. Some sophisticated noise-resistance counting methods have been proposed to alleviate this issue. Differently, we aim to directly refine the initial point annotations before training counting models. For that, we propose the Shifted Autoencoders (SAE), which enhances annotation consistency. Specifically, SAE applies random shifts to initial point annotations and employs a UNet to restore them to their original positions. Similar to MAE reconstruction, the trained SAE captures general position knowledge and ignores specific manual offset noise. This allows to restore the initial point annotations to more general and thus consistent positions. Extensive experiments show that using such refined consistent annotations to train some advanced (including noise-resistance) object counting models steadily/significantly boosts their performances. Remarkably, the proposed SAE helps to set new records on nine datasets. We will make codes and refined point annotations available. |
| title | Shifted Autoencoders for Point Annotation Restoration in Object Counting |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2312.07190 |