Computer Vision Perception Engineer

Role: Computer Vision Perception Engineer (Autonomous Driving)

Location:Detroit, MI (Fully onsite)

Job Description:

What You Will Do:

• Design and implement computer vision algorithms for object detection and segmentation using camera and LiDAR data fusion.
• Develop deep learning models for 2D and 3D object detection, including implementation and optimization of YOLO, Faster R-CNN, SSD, and transformer-based architectures.
• Create and optimize LiDAR point cloud processing pipelines using PCL and Open3D for 3D object detection and segmentation.
• Implement sensor fusion techniques to combine camera and LiDAR data for enhanced object detection accuracy.
• Develop instance and semantic segmentation algorithms using state-of-the-art models like Mask R-CNN, U-Net, and DeepLab.
• Implement and optimize deep learning models specifically designed for LiDAR point clouds, including PointNet, PointNet++, and other 3D neural network architectures.
• Develop robust perception algorithms that maintain performance in adverse weather conditions such as rain, snow, fog, and low-light scenarios.
• Build and maintain computer vision pipelines using OpenCV for image preprocessing, feature extraction, and geometric transformations.
• Design and implement multi-object tracking systems using Kalman filtering, SORT, and DeepSORT algorithms.
• Work with ROS2 for integration and deployment of perception algorithms.
• Optimize deep learning models for edge deployment and real-time inference performance.
• Develop robust evaluation metrics and testing frameworks for object detection systems.
• Collaborate with cross-functional teams to integrate perception algorithms into larger autonomous systems.
• Stay up-to-date with industry trends and emerging technologies to innovate and improve perception systems.

What You Will Bring:

• Strong expertise in computer vision and deep learning for object detection and segmentation tasks.
• Proficiency in deep learning frameworks (PyTorch and TensorFlow) with hands-on experience implementing detection models (YOLO, Faster R-CNN, SSD, RetinaNet, Detectron, etc.).
• Extensive experience with OpenCV for image processing and computer vision applications.
• Solid background in 3D perception using LiDAR point clouds; proficiency with PCL and Open3D libraries.
• Familiarity with LiDAR-specific deep learning models such as PointNet, PointNet++, VoxelNet, and other point cloud neural network architectures.
• Experience in developing and improving perception models for adverse weather conditions (rain, snow, fog) including domain adaptation and robust feature extraction techniques.
• Experience with sensor fusion techniques for combining camera and LiDAR data streams.
• Strong programming skills in Python and C++ for algorithm development and optimization.
• Experience with model optimization techniques for real-time inference.
• Familiarity with 3D geometry, coordinate transformations, and spatial data processing.
• Knowledge of evaluation metrics for object detection and tracking systems (mAP, IoU, custom metrics, etc.).