2018/04/03 by Kevan Yuen, Yuen, Kevan, Mohan M. Trivedi +1
Computer Science · Medicine · Psychology · #Computer Vision and Pattern Recognition (cs.CV) #Emotion and Mood Recognition #FOS: Computer and information sciences #FOS: Electrical engineering #Hand Gesture Recognition Systems #Image and Video Processing (eess.IV) #Musculoskeletal pain and rehabilitation #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.1804.01176
openalex publication_date 2018/04/03 · openalex created_date 2022/10/07 · openalex updated_date 2026/07/28
In the context of autonomous driving, where humans may need to take over in\nthe event where the computer may issue a takeover request, a key step towards\ndriving safety is the monitoring of the hands to ensure the driver is ready for\nsuch a request. This work, focuses on the first step of this process, which is\nto locate the hands. Such a system must work in real-time and under varying\nharsh lighting conditions. This paper introduces a fast ConvNet approach, based\non the work of original work of OpenPose for full body joint estimation. The\nnetwork is modified with fewer parameters and retrained using our own day-time\nnaturalistic autonomous driving dataset to estimate joint and affinity heatmaps\nfor driver & passenger's wrist and elbows, for a total of 8 joint classes and\npart affinity fields between each wrist-elbow pair. The approach runs real-time\non real-world data at 40 fps on multiple drivers and passengers. The system is\nextensively evaluated both quantitatively and qualitatively, showing at least\n95% detection performance on joint localization and arm-angle estimation.\n