To perform batched inference, the tensorrt engines must be compiled with a batch size of 2
To perform batched inference, the tensorrt engines must be compiled with a batch size of 2