Object Detection With Faster R-CNN in high resolution image

I am raising this question after doing efforts for about 2 days but getting no solution to my existing problem. I have an image of 5000 * 600 which i want to inference with Faster RCNN using tensorrt plugin.

What I read from the official documentation

Faster R-CNN takes 3 channel 375x500 images as input. I have seen the same dimension from the file in the line number 26 and 27. I have tried to change the dimensions in the line number 26 and 27 so that it can do inference. But such effort was not of any use. I am facing another issue. If I am going for inference of single image i.e using one image in imagelist. Please see line number 294 (std::vectorstd::string imageList = {“004545.ppm”, “test12.ppm”};), it shows the error as core dumped.

Which directory I am working

I have installed tensorrt in my ubuntu system using the official documentation of nvidia and is working in /usr/src/tensorrt where I have samples as well as data. Configuration file for FasterRCNN (sampleFasterRCNN.cpp) can be found in path /usr/src/tensorrt/samples/sampleFasterRCNN.

What efforts I have done to deal with core dumped

Instead of passing a single image, I have passed multiple image and such error is resolved. I did not understand why?

What I want to achieve

I want to customize the Faster RCNN with big and customize image using tensorrt plugin. Is there any way to do so. I am using the frozen file (.pb) from tensorflow zoo.

Please help me.

I hope readers and developers have understood my problem well. I am waiting for the solution. All ideas are welcomed.

Did you change the batch size below.
// Batch size
const int N = 5;