Mengdi Liu

Keliang Li, Yansong Li, Hongze Shen et al.

The dense, temporal nature of video presents a profound challenge for automated analysis. Despite the use of powerful Vision-Language Models, prevailing methods for video understanding are limited by the inherent disconnect between reasoning and perception: they rely on static, pre-processed information and cannot actively seek raw evidence from video as their understanding evolves. To address this, we introduce LensWalk, a flexible agentic framework that empowers a Large Language Model reasoner to control its own visual observation actively. LensWalk establishes a tight reason-plan-observe loop where the agent dynamically specifies, at each step, the temporal scope and sampling density of the video it observes. Using a suite of versatile, Vision-Language Model based tools parameterized by these specifications, the agent can perform broad scans for cues, focus on specific segments for fact extraction, and stitch evidence from multiple moments for holistic verification. This design allows for progressive, on-demand evidence gathering that directly serves the agent's evolving chain of thought. Without requiring any model fine-tuning, LensWalk delivers substantial, plug-and-play performance gains on multiple model recipes, boosting their accuracy by over 5\% on challenging long-video benchmarks like LVBench and Video-MME. Our analysis reveals that enabling an agent to control how it sees is key to unlocking more accurate, robust, and interpretable video reasoning.

Mengdi Liu, Qiang Li, Weizhi Nie et al.

Type A Aortic Dissection (TAAD) is a life-threatening cardiovascular emergency that demands rapid and precise preoperative evaluation. While key anatomical and pathological features are decisive for surgical planning, current research focuses predominantly on improving segmentation accuracy, leaving the reliable, quantitative extraction of clinically actionable features largely under-explored. Furthermore, constructing comprehensive TAAD datasets requires labor-intensive, expert level pixel-wise annotations, which is impractical for most clinical institutions. Due to significant domain shift, models trained on a single center dataset also suffer from severe performance degradation during cross-institutional deployment. This study addresses a clinically critical challenge: the accurate extraction of key TAAD clinical features during cross-institutional deployment in the total absence of target-domain annotations. To this end, we propose an unsupervised domain adaptation (UDA)-driven framework for the automated extraction of TAAD clinical features. The framework leverages limited source-domain labels while effectively adapting to unlabeled data from target domains. Tailored for real-world emergency workflows, our framework aims to achieve stable cross-institutional multi-class segmentation, reliable and quantifiable clinical feature extraction, and practical deployability independent of high-cost annotations. Extensive experiments demonstrate that our method significantly improves cross-domain segmentation performance compared to existing state-of-the-art approaches. More importantly, a reader study involving multiple cardiovascular surgeons confirms that the automatically extracted clinical features provide meaningful assistance for preoperative assessment, highlighting the practical utility of the proposed end-to-end segmentation-to-feature pipeline.

Qibo Chen, Weizhong Jin, Jianyue Ge et al.

Recent research on universal object detection aims to introduce language in a SoTA closed-set detector and then generalize the open-set concepts by constructing large-scale (text-region) datasets for training. However, these methods face two main challenges: (i) how to efficiently use the prior information in the prompts to genericise objects and (ii) how to reduce alignment bias in the downstream tasks, both leading to sub-optimal performance in some scenarios beyond pre-training. To address these challenges, we propose a strong universal detection foundation model called CP-DETR, which is competitive in almost all scenarios, with only one pre-training weight. Specifically, we design an efficient prompt visual hybrid encoder that enhances the information interaction between prompt and visual through scale-by-scale and multi-scale fusion modules. Then, the hybrid encoder is facilitated to fully utilize the prompted information by prompt multi-label loss and auxiliary detection head. In addition to text prompts, we have designed two practical concept prompt generation methods, visual prompt and optimized prompt, to extract abstract concepts through concrete visual examples and stably reduce alignment bias in downstream tasks. With these effective designs, CP-DETR demonstrates superior universal detection performance in a broad spectrum of scenarios. For example, our Swin-T backbone model achieves 47.6 zero-shot AP on LVIS, and the Swin-L backbone model achieves 32.2 zero-shot AP on ODinW35. Furthermore, our visual prompt generation method achieves 68.4 AP on COCO val by interactive detection, and the optimized prompt achieves 73.1 fully-shot AP on ODinW13.

Qibo Chen, Weizhong Jin, Shuchang Li et al.

Text prompts are crucial for generalizing pre-trained open-set object detection models to new categories. However, current methods for text prompts are limited as they require manual feedback when generalizing to new categories, which restricts their ability to model complex scenes, often leading to incorrect detection results. To address this limitation, we propose a novel visual prompt method that learns new category knowledge from a few labeled images, which generalizes the pre-trained detection model to the new category. To allow visual prompts to represent new categories adequately, we propose a statistical-based prompt construction module that is not limited by predefined vocabulary lengths, thus allowing more vectors to be used when representing categories. We further utilize the category dictionaries in the pre-training dataset to design task-specific similarity dictionaries, which make visual prompts more discriminative. We evaluate the method on the ODinW dataset and show that it outperforms existing prompt learning methods and performs more consistently in combinatorial inference.