Search papers, labs, and topics across Lattice.
This paper introduces Discriminative Evidence Composition (DeCO), a novel approach to fine-grained dataset distillation that focuses on preserving localized evidence critical for visual classification tasks. By leveraging attention rollout from a pretrained TransFG teacher, DeCO identifies and organizes informative image patches into class-wise evidence banks, reducing redundancy through spatial diversification. Experimental results demonstrate that DeCO significantly outperforms existing coreset and dataset-distillation methods across multiple fine-grained datasets, highlighting its effectiveness in maintaining classification performance with compact synthetic sets.
Fine-grained dataset distillation can achieve superior performance by focusing on localized evidence rather than just global statistics.
Dataset distillation compresses a large training set into a compact synthetic set while preserving its downstream utility. However, existing methods primarily preserve global image statistics and may overlook the localized evidence essential for fine-grained visual classification (FGVC), such as object parts, subtle textures, and region-specific structures. We formulate fine-grained dataset distillation as budgeted discriminative-evidence preservation and propose Discriminative Evidence Composition (DeCO). DeCO uses attention rollout from a pretrained TransFG teacher to identify informative patches, applies spatial diversification to reduce redundant coverage, and organizes the resulting regions into class-wise evidence banks. Multiple same-class regions are then packed into compact grid-composed images. The teacher is used only for dataset construction, whereas downstream students are trained with standard hard-label supervision without teacher logits. Experiments on CUB-200-2011, FGVC-Aircraft, and Stanford Cars show that DeCO consistently outperforms representative coreset and dataset-distillation baselines under different IPC budgets.
