About
Embedded machine learning is becoming increasingly important, and so are the techniques for optimizing and lightweighting deep neural networks. We are conducting research on computer vision for autonomous driving. For autonomous driving, it is essential to run high-performance algorithms in real time under hardware constraints. To implement a real-time object detection algorithm based on multi-modal sensors, we are developing methods for processing and fusing various sensor data, such as images and LiDAR, as well as optimizing and lightweighting CNN and transformer architectures. Furthermore, we are researching Neural Architecture Search to find the optimal network structure for specific hardware
