How do I prepare '_boxes.mat'file for my own datasets?

[jojotata]

Hello. I want to use fast R-CNN for my research. I already run demo.py successfully. So, I tried to run this using my own datasets. However, when I tried this, I got the error message "No such file 'sample_boxes.mat'". (my own dataset's filename was "sample.jpg")

My question is how do I prepare '_boxes.mat' file for my own datasets?

Sorry to ask you such a basic question.

[catsdogone]

hi. You can use 'selective search' or other methods for generating region proposals.

[frattezi]

I'm using the python wrapper described on the readme, you may have to change part of the code. On selective_search. py comment line 46 : #os.remove(output_filename) for keeping the _boxes.m on your tmp files (you can also change the final location of the files). You may have to change the line 40 on the selective_search.m to : save(output_filename, 'boxes', '-v7');. That worked to me, good luck!

[kshalini]

@frattezi do you have a complete set of steps that you can outline to help with the training? i am facing the same problem, but not able to find any clear set of steps to get through from end to end, ie what configs and code changes to make, to enable training on my custom dataset. thanks.

[arash111]

anybody can help please!! I have the same problem and don't know how to prepare '_boxes.mat'file for my own datasets in matlab. I'm using the code proposed here: (https://github.com/vlfeat/matconvnet/blob/master/examples/fast_rcnn/fast_rcnn_demo.m). thanks,