Frequency-Adaptive Discrete Cosine-ViT-ResNet Architecture for Sparse-Data Vision
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866910178165653504 |
|---|---|
| author | Kang, Ziyue Zhang, Weichuan |
| author_facet | Kang, Ziyue Zhang, Weichuan |
| contents | A major challenge in rare animal image classification is the scarcity of data, as many species usually have only a small number of labeled samples.
To address this challenge, we designed a hybrid deep-learning framework comprising a novel adaptive DCT preprocessing module, ViT-B16 and ResNet50 backbones, and a Bayesian linear classification head. To our knowledge, we are the first to introduce an adaptive frequency-domain selection mechanism that learns optimal low-, mid-, and high-frequency boundaries suited to the subsequent backbones.
Our network first captures image frequency-domain cues via this adaptive DCT partitioning. The adaptively filtered frequency features are then fed into ViT-B16 to model global contextual relationships, while ResNet50 concurrently extracts local, multi-scale spatial representations from the original image. A cross-level fusion strategy seamlessly integrates these frequency- and spatial-domain embeddings, and the fused features are passed through a Bayesian linear classifier to output the final category predictions. On our self-built 50-class wildlife dataset, this approach outperforms conventional CNN and fixed-band DCT pipelines, achieving state-of-the-art accuracy under extreme sample scarcity. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_22701 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Frequency-Adaptive Discrete Cosine-ViT-ResNet Architecture for Sparse-Data Vision Kang, Ziyue Zhang, Weichuan Computer Vision and Pattern Recognition A major challenge in rare animal image classification is the scarcity of data, as many species usually have only a small number of labeled samples. To address this challenge, we designed a hybrid deep-learning framework comprising a novel adaptive DCT preprocessing module, ViT-B16 and ResNet50 backbones, and a Bayesian linear classification head. To our knowledge, we are the first to introduce an adaptive frequency-domain selection mechanism that learns optimal low-, mid-, and high-frequency boundaries suited to the subsequent backbones. Our network first captures image frequency-domain cues via this adaptive DCT partitioning. The adaptively filtered frequency features are then fed into ViT-B16 to model global contextual relationships, while ResNet50 concurrently extracts local, multi-scale spatial representations from the original image. A cross-level fusion strategy seamlessly integrates these frequency- and spatial-domain embeddings, and the fused features are passed through a Bayesian linear classifier to output the final category predictions. On our self-built 50-class wildlife dataset, this approach outperforms conventional CNN and fixed-band DCT pipelines, achieving state-of-the-art accuracy under extreme sample scarcity. |
| title | Frequency-Adaptive Discrete Cosine-ViT-ResNet Architecture for Sparse-Data Vision |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2505.22701 |