Training Take Long time

Hi Rajendra

On the picture my setup inside my DLS , I want to do testing for the Images inside (test) subfolder (home\1\dataset\test

I zipped (test) then I will upload it, for the csv I wrote each line the folder address text (I assumed csv inside test folder) as the following format : “./img_XXXX.jpg” , Is the processes correct?

yes. that is correct. I am assuming your structure is as follows

test
|
--------test.csv
|
--------img_XXXX.jpg

Ok , But when I upload the file it failed may be because I zipped the test folder (3 GB) , shall I zip only the csv file or the csv+the Images inside (test) folder?

your test dataset needs to have CSV + Images. There is limit of 1GB of on the upload size on the inference tab.

Either you can divide it into smaller datasets or I can suggest a workaround.

Workaround is to have full path /data/1/datasets/test/img_XXXX.jpg in the test.csv and then you can upload without the image files in the zip since images are already there inside /data/1/datasets/test.

That exactly what I did I Zipped only the csv but I got an error

Did you zip test.csv or the folder containing test.csv ? You need to zip the folder containing test.csv file.

I mean the workaround , I did the work around

I wrote on the csv file (zipped) : ./data/1/datasets/test/img_XXXX.jpg

If you start a path with ., it is treated as relative path. Remove . from your pathname (e.g. /data/1/datasets/test/img_XXXX.jpg)

Still Not clear could you explain more?

The path should be /data/1/datasets/test/img_XXXX.jpg (notice that leading . has been removed in this path)

1 Like

As Per your Recommendation I did it but still display message “No such a file or directory” , I attached all the needed screen-shots.

You are right. We are able to reproduce this issue.

You need to use file path as ../../../datasets/test/img_XXXX.jpg

No Still The same message and the same issue :rage:

Can you show few lines of your test.csv?

Here is a screen-shot , this is as per your recommendation to put (…/…/…/)

This looks fine to me. I am not sure why this does not work in your environment. Here is my test.csv snapshot which is working fine with our windows version.

If you feel like debugging further. You can launch bash.exe as shown in the following snapshot. Run the export command to set the PATH. Go to the test folder of your project (replace project id 0 with your project id). Check if the path mentioned in your test.csv file is accessible or not.

I have anaconda for windows but i don’t know how to reach it can you guide me how to go into anaconda prompt?

Like the one on your window?

My Path On Windows :slight_smile:

C:\DLS\DeepLearningStudio\data\1\datasets\test