Utilizing thermal data for designing efficient AI-based object detection and classification pipeline for Advance Driver-Assistance Systems.
Abstract
“This study is focused on evaluating the real-time performance of thermal object detection for smart and safe vehicular systems by deploying the trained networks on GPU & single-board EDGE-GPU computing platforms for onboard automotive sensor suite testing. A novel large-scale thermal dataset comprising of > 35,000 distinct frames is acquired, processed, and open-sourced in challenging weather and environmental scenarios. The dataset is a recorded from lost-cost yet effective uncooled LWIR thermal camera, mounted stand-alone and on an electric vehicle to minimize mechanical vibrations. State-of-the-art YOLO-V5 networks variants are trained using four different public datasets as well newly acquired local dataset for optimal generalization of DNN by employing SGD optimizer. The effectiveness of trained networks is validated on extensive test data using various quantitative metrics which include precision, recall curve, mean average precision, and frames per second. The smaller network variant of YOLO is further optimized using TensorRT inference accelerator to explicitly boost the frames per second rate. Optimized network engine increases the frames per second rate by 3.5 times when testing on low power edge devices thus achieving 11 fps on Nvidia Jetson Nano and 60 fps on Nvidia Xavier NX development boards.”
Find the technical paper here and here.
Muhammad Ali Farooq, Waseem Shariff, Peter Corcoran. Submitted Jan 2022.
Sensor technologies are still evolving, and capabilities are being debated.
Academia, industry partnerships ramp to entice undergrads into hardware engineering.
Pitches continue to decrease, but new tooling and technologies are required.
Buried features and re-entrant geometries drive application-specific metrology solutions.
Issues involving design, manufacturing, packaging, and observability all need to be solved before this approach goes mainstream for many applications.
While terms often are used interchangeably, they are very different technologies with different challenges.
Technology and business issues mean it won’t replace EUV, but photonics, biotech and other markets provide plenty of room for growth.
Commercial chiplet marketplaces are still on the distant horizon, but companies are getting an early start with more limited partnerships.
Existing tools can be used for RISC-V, but they may not be the most effective or efficient. What else is needed?
How customization, complexity, and geopolitical tensions are upending the global status quo.
Sensor technologies are still evolving, and capabilities are being debated.
The industry is gaining ground in understanding how aging affects reliability, but more variables make it harder to fix.
Key pivot and innovation points in semiconductor manufacturing.
Leave a Reply