# Saccadic Vision for Fine-Grained Visual Classification

arXiv: https://arxiv.org/abs/2509.15688 · HTML: https://mlsanity.com/paper/2509.15688/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 40 of its 62 references that are in the map.
- First posted on arXiv 19 Sep 2025.

## Builds on (most cited first, 30 of 40)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://mlsanity.com/paper/2103.14030/) (2021)
- [Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks](https://mlsanity.com/paper/1506.01497/) (2015)
- [Mask R-CNN](https://mlsanity.com/paper/1703.06870/) (2017)
- [Layer Normalization](https://mlsanity.com/paper/1607.06450/) (2016)
- [Feature Pyramid Networks for Object Detection](https://mlsanity.com/paper/1612.03144/) (2016)
- [You Only Look Once: Unified, Real-Time Object Detection](https://mlsanity.com/paper/1506.02640/) (2015)
- [Squeeze-and-Excitation Networks](https://mlsanity.com/paper/1709.01507/) (2017)
- [Training data-efficient image transformers & distillation through attention](https://mlsanity.com/paper/2012.12877/) (2020)
- [A ConvNet for the 2020s](https://mlsanity.com/paper/2201.03545/) (2022)
- [Gaussian Error Linear Units (GELUs)](https://mlsanity.com/paper/1606.08415/) (2016)
- [SSD: Single Shot MultiBox Detector](https://mlsanity.com/paper/1512.02325/) (2015)
- [Fast R-CNN](https://mlsanity.com/paper/1504.08083/) (2015)
- [Aggregated Residual Transformations for Deep Neural Networks](https://mlsanity.com/paper/1611.05431/) (2016)
- [Rich feature hierarchies for accurate object detection and semantic segmentation](https://mlsanity.com/paper/1311.2524/) (2013)
- [DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification](https://mlsanity.com/paper/2106.02034/) (2021)
- [ImageNet-21K Pretraining for the Masses](https://mlsanity.com/paper/2104.10972/) (2021)
- [Continual Test-Time Domain Adaptation](https://mlsanity.com/paper/2203.13591/) (2022)
- [AdaViT: Adaptive Tokens for Efficient Vision Transformer](https://mlsanity.com/paper/2112.07658/) (2021)
- [Multi-Scale Vision Longformer: A New Vision Transformer for High-Resolution Image Encoding](https://mlsanity.com/paper/2103.15358/) (2021)
- [TransFG: A Transformer Architecture for Fine-grained Recognition](https://mlsanity.com/paper/2103.07976/) (2021)
- [Part-Stacked CNN for Fine-Grained Visual Categorization](https://mlsanity.com/paper/1512.08086/) (2015)
- [Test-Time Domain Adaptation by Learning Domain-Aware Batch Normalization](https://mlsanity.com/paper/2312.10165/) (2023)
- [Inverse Compositional Spatial Transformer Networks](https://mlsanity.com/paper/1612.03897/) (2016)
- [Learning for Single-Shot Confidence Calibration in Deep Neural Networks through Stochastic Inferences](https://mlsanity.com/paper/1809.10877/) (2018)
- [Fine-grained Visual Classification with High-temperature Refinement and Background Suppression](https://mlsanity.com/paper/2303.06442/) (2023)
- [Advancing Fine-Grained Classification by Structure and Subject Preserving Augmentation](https://mlsanity.com/paper/2406.14551/) (2024)
- [A Novel Plug-in Module for Fine-Grained Visual Classification](https://mlsanity.com/paper/2202.03822/) (2022)
- [Probabilistic Spatial Transformer Networks](https://mlsanity.com/paper/2004.03637/) (2020)

## Built on by (most cited first, 2 of 2)

- [Beyond Scalar Distances: Semantic Attribute Gradients from Frozen MLLMs for Visual Embeddings](https://mlsanity.com/paper/2606.15134/) (2026)
- [Zero-Shot Test-Time Canonicalization using Out-of-Distribution Scoring](https://mlsanity.com/paper/2606.24178/) (2026)

## Abstract (arXiv)

Fine-grained visual classification (FGVC) requires distinguishing between visually similar categories through subtle, localized features - a task that remains challenging due to high intra-class variability and limited inter-class differences. Existing part-based methods often rely on complex localization networks that learn mappings from pixel to sample space, requiring a deep understanding of image content while limiting feature utility for downstream tasks. In addition, sampled points frequently suffer from high spatial redundancy, making it difficult to quantify the optimal number of required parts. Inspired by human saccadic vision, we propose a two-stage process that first extracts peripheral features (coarse view) and generates a sample map, from which fixation patches are sampled and encoded in parallel using a weight-shared encoder. We employ contextualized selective attention to weigh the impact of each fixation patch before fusing peripheral and focus representations. To prevent spatial collapse - a common issue in part-based methods - we utilize non-maximum suppression during fixation sampling to eliminate redundancy. Comprehensive evaluation on standard FGVC benchmarks (CUB-200-2011, NABirds, Food-101 and Stanford-Dogs) and challenging insect datasets (EU-Moths, Ecuador-Moths and AMI-Moths) demonstrates that our method achieves comparable performance to state-of-the-art approaches while consistently outperforming our baseline encoder.