Introduction

Kutuma is used for file uploads to cloud, but it can also be used to transfer data from SD Card to the external disk. This can be done through the Settings menu of the Kutuma app.

Step 1: Sign in to Kutuma

Go through the normal sign in steps of Kutuma, including past the ArcGIS Online (AGOL) sign in page.

Step 2: Check the default Kutuma Storage

Go to the Settings menu on the lower left of the Kutuma inteface. You will see the default storage location for Kutuma under Preferences>Current Storage

Step 3: Change the default Kutuma Storage

Click the Change the storage location button and select the path in your external drive where you would like to transfer the files from the sensor SD Card to now in the external drive where you will run the classifiers. For example, in the below image the default location has been changed to D:>test_uploads.

Step 4: Proceed with Data Import

Now that you have changed the default storage location to the one in your external disk, you can proceed with the data import steps. Once you finish Step 1: SD card detection, Step 2: SD Card connection, Step 3: Data type selection and Step 4: Data Import and Fine Tuning, at the end of step 4 your SD card files will be transfered to the new location in your external drive.

Step 5: Check if files exist in the external drive

Go to your external drive where you will be running the ML Classifiers, from. For example, the external drive where I will be running the classifiers is Drive D:. Since I had changed my default Kutuma storage location to D:>test_uploads, I will go to my test_uploads folder inside my external disk.

Inside this test_uploads folder, I will find that Kutuma has already created a folder called uploads>Natural_State>XPRIZE>cameraTrapImages. The preultimate folder is cameraTrapImages because that is the data type I selected in Step 3. Inside that folder is a folder named with QR Code of the SD Card I scanned with using Kutuma’s camera.

Here are the files inside the QR code named folder inside my external drive.

Step 6: Run the classifier on the external drive

If you see above, the default storage location has been changed to: <folder-path>/cameraTrapImages. It can be any folder path/folder-name depending on the new storage location path you defined in Kutuma and the data type selected.

For this section let’s assume we selected audio files, already transfered the audio files from the SD card to the external drive. The files are saved within this folder path /mnt/d/test_uploads/uploads/Natural_State/XPRIZE/audioFiles.

When you are running docker, the path from which docker will read the files is -v /mnt/d/test_uploads/uploads/Natural_State/XPRIZE/audioFiles:/app/input \. Note that this is the path in my external disk.

However, it can save the outputs in any path you specify. For example, I specify my outputs to be in the -v /mnt/d/test_classifiers/birdnet_amphReptiles_test/birdnet_processed:/app/birdnet_processed \, and my detections to the -v /mnt/d/test_classifiers/birdnet_amphReptiles_test/tmp_prediction:/app/tmp_prediction \ and so on. My species list can also be read from anywhere provided it is the right species list. For example, this is the path to my species list: -v /mnt/d/test_classifiers/birdnet_amphReptiles_test/species_list.

Assuming we want to run the birdnet amphiReptiles classifier on the data transfered to the path ``/mnt/d/test_uploads/uploads/Natural_State/XPRIZE/audioFiles`, this is how our docker run command will look like.

docker run \
    --shm-size=10.24gb \
    -v /mnt/d/test_uploads/uploads/Natural_State/XPRIZE/audioFiles:/app/input \
    -v /mnt/d/test_classifiers/birdnet_amphReptiles_test/birdnet_processed:/app/birdnet_processed \
    -v /mnt/d/test_classifiers/birdnet_amphReptiles_test/tmp_prediction:/app/tmp_prediction \
    -v /mnt/d/test_classifiers/birdnet_amphReptiles_test/detections:/app/detections \
    -v /mnt/d/test_classifiers/birdnet_amphReptiles_test/species_list:/app/slist \
    naturalstate/birdnet_amph_reptiles:latest \
    --phase_nb 0 \
    --acoustic_test_data /app/input \
    --slist /app/slist