Frequency-Adaptive Discrete Cosine-ViT-ResNet Architecture for Sparse-Data Vision
This work addresses the problem of data scarcity in rare animal classification for conservationists and researchers, though it is incremental as it builds on existing ViT and ResNet architectures.
The paper tackled rare animal image classification with scarce data by designing a hybrid deep-learning framework that adaptively selects frequency-domain features and fuses them with spatial representations, achieving state-of-the-art accuracy on a 50-class wildlife dataset under extreme sample scarcity.
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.