In TensorRT, when I run the engine with googlenet model:
How to create the buffers array with 2 images when batchSize set to 2, and return an array with the classification results of the 2 images?
Does anyone have an example code?
Multiple input idea comes from training.
In training, to accelerate the update speed, users usually input multiple images and get the prediction at the same time.
Similar, you can put multiple images with batch size > 1 and have the prediction results of all the given data with one execute() call.
Ex. batch size = 8
Input = 8xHxWxC
Output = 8xClass