Search papers, labs, and topics across Lattice.
The paper introduces UniDD, a universal dataset distillation framework based on a task-driven diffusion model, to address the limitations of existing dataset distillation methods that primarily focus on image classification and struggle with task optimization heterogeneity and inflexible image generation for detection and segmentation tasks. UniDD employs a two-stage approach: Universal Task Knowledge Mining using task-specific proxy models, followed by Universal Task-Driven Diffusion guided by these proxies to generate task-specific synthetic images. Experiments on ImageNet-1K, Pascal VOC, and MS COCO demonstrate that UniDD outperforms state-of-the-art methods, achieving a 6.1% improvement over previous diffusion-based methods on ImageNet-1K with IPC-10 while reducing deployment costs.
Dataset distillation that works well for image classification, object detection, and segmentation is now possible thanks to a new diffusion-based approach.
Dataset distillation (DD) condenses key information from large-scale datasets into smaller synthetic datasets, reducing storage and computational costs for training networks. However, most recent research has primarily focused on image classification tasks, with limited exploration in detection and segmentation. Two key challenges remain: (i) Task Optimization Heterogeneity, where existing methods focus on class-level information but fail to address the diverse needs of detection and segmentation, and (ii) Inflexible Image Generation, where current generation methods rely on global updates for single-class targets and lack localized optimization for specific object regions. To address these challenges, we propose UniDD, a universal dataset distillation framework built on a task-driven diffusion model for diverse DD tasks, as shown in Fig. 1. Our approach operates in two stages: Universal Task Knowledge Mining, which captures task-relevant information through task-specific proxy model training, and Universal Task-Driven Diffusion, where these proxies guide the diffusion process to generate task-specific synthetic images. Extensive experiments across ImageNet-1K, Pascal VOC, and MS COCO demonstrate that UniDD consistently outperforms state-of-the-art methods. In particular, on ImageNet-1K with IPC-10, UniDD surpasses previous diffusion-based methods by 6.1%, while also reducing deployment costs.
