Development of Low-Cost Real-Time Driver Drowsiness Detection System using Eye Centre Tracking and Dynamic Thresholding
为减少因驾驶员疲劳引起的交通事故,开发了一种基于眼中心跟踪和动态阈值处理的低成本实时驾驶员困倦检测系统。
为减少因驾驶员疲劳引起的交通事故,开发了一种基于眼中心跟踪和动态阈值处理的低成本实时驾驶员困倦检测系统。
为解决冷冻电镜断层扫描中子断层图分类因标注样本稀缺而面临的难题,提出了一种新的合成到真实的适应框架,通过可学习的转换模块在输入和特征层面缩小合成与真实数据间的差距。
本文研究了图中有效k-限制广播支配问题,通过动态规划算法为树状图提供多项式时间解法,并证明该问题是NP完全的。
This work addresses the limited generalization of existing deepfake detection methods in the face of rapidly evolving generative models and diverse forgery techniques. It proposes a robust detection system by integrating fine-tuned state-of-the-art vision Transformer models—DINOv2, AIMv2, and OpenCLIP ViT-L/14—and optimizes them on the large-scale in-the-wild dataset DF-Wild. Evaluated on the DF-Wild test set, the proposed approach achieves an AUC of 96.77% and an EER of 9%, outperforming the current best method by 7.05% in AUC and 8% in EER. This significant improvement in detecting unseen forgery types earned the method first place in the IEEE S&P Cup 2025.
Current automated methods for retinal cyst segmentation exhibit limited accuracy (only 68%) and insufficient robustness on high-noise OCT images—particularly those acquired with Topcon devices—hindering their clinical utility for precise quantification. To address this, this work proposes a ResNet-based patch classification strategy and conducts training and evaluation on a publicly available challenge dataset encompassing multi-vendor imaging systems and annotations from multiple experts. It presents the first systematic analysis of generalization performance across different OCT devices for cyst segmentation. The proposed method achieves Dice scores exceeding 70% across all vendors, significantly outperforming existing state-of-the-art approaches and demonstrating markedly improved segmentation accuracy and robustness to variations in image quality.
为减少因驾驶员疲劳引起的交通事故,开发了一种基于眼中心跟踪和动态阈值处理的低成本实时驾驶员困倦检测系统。
为解决冷冻电镜断层扫描中子断层图分类因标注样本稀缺而面临的难题,提出了一种新的合成到真实的适应框架,通过可学习的转换模块在输入和特征层面缩小合成与真实数据间的差距。
本文研究了图中有效k-限制广播支配问题,通过动态规划算法为树状图提供多项式时间解法,并证明该问题是NP完全的。
This work addresses the limited generalization of existing deepfake detection methods in the face of rapidly evolving generative models and diverse forgery techniques. It proposes a robust detection system by integrating fine-tuned state-of-the-art vision Transformer models—DINOv2, AIMv2, and OpenCLIP ViT-L/14—and optimizes them on the large-scale in-the-wild dataset DF-Wild. Evaluated on the DF-Wild test set, the proposed approach achieves an AUC of 96.77% and an EER of 9%, outperforming the current best method by 7.05% in AUC and 8% in EER. This significant improvement in detecting unseen forgery types earned the method first place in the IEEE S&P Cup 2025.
Current automated methods for retinal cyst segmentation exhibit limited accuracy (only 68%) and insufficient robustness on high-noise OCT images—particularly those acquired with Topcon devices—hindering their clinical utility for precise quantification. To address this, this work proposes a ResNet-based patch classification strategy and conducts training and evaluation on a publicly available challenge dataset encompassing multi-vendor imaging systems and annotations from multiple experts. It presents the first systematic analysis of generalization performance across different OCT devices for cyst segmentation. The proposed method achieves Dice scores exceeding 70% across all vendors, significantly outperforming existing state-of-the-art approaches and demonstrating markedly improved segmentation accuracy and robustness to variations in image quality.