Gesture-Based Control System in Automotive
summary
Camera-based gesture control for cars is implemented using an existing driver-monitoring camera and NVIDIA Jetson Nano running DeepStream and TensorRT. The system classifies gestures with an MLP trained on a custom dataset (standardized start/stop gestures and cyclic gestures to cut false positives), achieves about 40 fps and uses ~650 MB on the board, works in simulator and basic in-car tests, but still needs roughly 10% more accuracy for production reliability.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 5:44Distracted driving causes
- 11:47Existing gesture solutions
- 16:09Gesture selection and data
- 21:31Simulation and real testing
- 24:58Hardware choice and optimization
- 33:26Q&A
related talks
13mNFCGest: Gesture Interaction with NFC terminal Enabled by Super High Speed ADC
40mSecuring Tesla and Mobileye from Split-Second Phantom Attacks
38mHow We Built a Safety Violations Detection System
16mCANdid - An Open-Access Annotated Dataset of Vehicle CAN Bus Traffic
18mModels on the Move: Towards Feasible Embedded AI for Intrusion Detection on...
44mConnected vehicle security: An AESIN tech talk on ADAS (advanced driver assistance systems)
01h 06mSelf-Driving and Connected Cars: Fooling Sensors and Tracking Drivers
50mSelf-driving cars - the future is here....