The Ipsotek VISuite video processing nodes require a GPU to perform AI inferencing on video from the camera streams. Three parameters can be tuned to increase the processing scale:
- The resource configuration of the processing node VMs
- The model and number of GPUs
- The number of load-balanced processing VMs.
We kept the resource allocation for all the VMs consistent with previous validations using A40 GPUs as documented in the published validation reports hosted on the Dell Infohub website - Virtualized Computer Vision for Smart Transportation platform.