8 papers
VideoResearcher: Self-Improving Tool Design for Long-Video Understanding
Dingqiang Ye, Dongdi Zhao, Kaishen Wang +11
Video agents have made substantial progress in long-video understanding. Yet effective video-agent systems require costly, time-consuming manual design and trial and error. Current…
Thermal-Det: Language-Guided Cross-Modal Distillation for Open-Vocabulary Thermal Object Detection
Yasiru Ranasinghe, Elim Schenck, Florence Yellin +3
Existing open-vocabulary detectors focus on RGB images and fail to generalize to thermal imagery, where low texture and emissivity variations challenge RGB-based semantics. We pres…
On-Policy Distillation with Best-of-N Teacher Rollout Selection
Ke Zhang, Yunjie Tian, Dongdi Zhao +4
On-policy distillation (OPD), which supervises a student on its own sampled trajectories, has emerged as a data-efficient post-training method for improving reasoning while avoidin…
On the structure of categorical duality operators
Corey Jones, Xinping Yang
We systematically study categorical duality operators on spin (and anyon) chains with respect to an internal fusion category symmetry C. We parameterize duality operators on the qu…
ProCrop: Learning Aesthetic Image Cropping from Professional Compositions
Ke Zhang, Tianyu Ding, Jiachen Jiang +4
Image cropping is crucial for enhancing the visual appeal and narrative impact of photographs, yet existing rule-based and data-driven approaches often lack diversity or require an…
HarmonySeg: Tubular Structure Segmentation with Deep-Shallow Feature Fusion and Growth-Suppression Balanced Loss
Yi Huang, Ke Zhang, Wei Liu +6
Accurate segmentation of tubular structures in medical images, such as vessels and airway trees, is crucial for computer-aided diagnosis, radiotherapy, and surgical planning. Howev…
MedCL: Learning Consistent Anatomy Distribution for Scribble-supervised Medical Image Segmentation
Ke Zhang, Vishal M. Patel
Curating large-scale fully annotated datasets is expensive, laborious, and cumbersome, especially for medical images. Several methods have been proposed in the literature that make…
ModelMix: A New Model-Mixup Strategy to Minimize Vicinal Risk across Tasks for Few-scribble based Cardiac Segmentation
Ke Zhang, Vishal M. Patel
Pixel-level dense labeling is both resource-intensive and time-consuming, whereas weak labels such as scribble present a more feasible alternative to full annotations. However, tra…