We are writing to ask for your suggestion on the choice of Texas Instruments processors for Automotive Embedded Vision application. We need to run the following algorithms on the processor from the images captured through a stereo camera running at 30 fps and 2 Megapixel quality.
-
Image pre-processing: Lens distortion, sensor tilting, gamma correction, RBG to greyscale conversion
-
Foreground Segmentation: Hypothesis generation and Verification using template and appearance based methods
-
Object Detection & Classification: YOLO, R-CNN, and machine learning classifiers
-
Object Tracking
-
Target Distance and Speed Estimation: Using stereo camera
Above mentioned algorithms require massive parallel processing. Therefore, a CPU executing multiple threads with 2 to 8 GB RAM along with fast GPU used in Autonomous vehicles, capable of producing real time outputs with minimal latency is the key. Furthermore, for quick image acquisition, a board compatible with MIPI CSI-2 camera interface that can support atleast 2 cameras while maintaining the desired resolution is also required.
Thank you in advance for your help. Please contact us if you need any further information.
