Advancements in Computer Vision for UAV and Roadway Safety

The field of computer vision is moving towards more integrated and multimodal approaches to improve safety in various applications, including UAV target detection and roadway obstacle detection. Researchers are exploring the use of combined sensor data, such as RGB, depth, and infrared images, to enhance detection accuracy and robustness. Additionally, there is a growing interest in using event-based cameras, which can capture high-quality images in diverse lighting conditions, for applications like civil infrastructure defect detection. Notably, innovative frameworks and models are being developed to address challenges like scale variations, class imbalance, and nonuniformity correction in UAV images. These advancements have the potential to significantly improve safety and efficiency in various industries. Noteworthy papers include: AD-Det, which proposes a novel framework for boosting object detection in UAV images with focused small objects and balanced tail classes, achieving state-of-the-art performance on public datasets. The ev-CIVIL dataset and benchmark, which introduces the first event-based civil infrastructure defect detection dataset, enabling accurate defect detection and classification even under challenging lighting conditions.

Sources

Real-Time Roadway Obstacle Detection for Electric Scooters Using Deep Learning and Multi-Sensor Fusion

Detection-Friendly Nonuniformity Correction: A Union Framework for Infrared UAVTarget Detection

AD-Det: Boosting Object Detection in UAV Images with Focused Small Objects and Balanced Tail Classes

Event-based Civil Infrastructure Visual Defect Detection: ev-CIVIL Dataset and Benchmark

Built with on top of