SemiField Preprocessing

TODOs (updated 12/10/2024)

1. Minor cleaning/formatting for both Field and SemiF

2. Flow chart of data transfer and processing for both Field and SemiF (Matthew Kutugata) could be a general flowchart placed in the main preprocessing page.

   1. Courtney Belcher Courtney Belcher could give it a shot

Backlog Spreadsheet:

1. Download Batch

When downloading a Semifield batch, It copies over to the local user from the NFS Lockers.

cd ..
cd psa_images/semifield_tools/
screen -R <batch_name>
python3 copy_from_lockers.py <batch_name>

Semifield batches can take anywhere from a few minutes to half an hour to copy over, depending on the batch size. It is important to open a new screen for copying. Only to copy one or two batches at a time within their respective screens to avoid disk space problems.

To exit the screen, use CTRL+A D

Checking in to see if the copying of a batch has been completed by using this string of commands:

cd psa_images/semifield_tools/
screen -r <batch_name>

If the batch has either been completed copying or there has been an error copying, there will be an input line. If no input line, the process has not completed.

To exit the screen, use CTRL+A D

1.1 Checking if the batch was completely copied

It is important to check if the copied batch was brought to the local user entirely.

We must know the # of images in the original batch. Use this command:

ls /mnt/research-projects/s/screberg/longterm_images2/semifield-upload/<batch>/*.ARW | wc -l

Check the # of images in the local directory after download has completed.

(Moving files from longterm storage to the local working directory, before you run the the pipeline).

*The output should be equal to the above command; should be 0 after running the pipeline

Use this command:

ls /home/psa_images/temp_data/semifield-upload/<batch>/*.ARW | wc -l

After running the pipeline, check if all the ARW files have converted to JPGs and have moved to the longterm “developed” storage.

Use this command:

ls /mnt/research-projects/s/screberg/longterm_images2/semifield-developed-images/<batch>/images/*.jpg | wc -l

 ls /mnt/research-projects/s/screberg/longterm_images2/semifield-upload/MD_2024-07-02/*.ARW | wc -l

ls /home/psa_images/temp_data/semifield-upload/MD_2024-07-02/*.ARW | wc -l

ls /mnt/research-projects/s/screberg/longterm_images2/semifield-developed-images/MD_2024-07-02/images/*.jpg | wc -l

2. Color Profiling in RawTherapee

Launch RawTherapee:

./RawTherapee_5.8.AppImage

Be sure to be located within the Semifield_tools folder.

RawTherapee will launch in a separate window, opening to the most recent profile that has been worked on. Select the appropriate batch and allow to load. Scroll through, making a mental note if some images are out of the ordinary (i.e. too bright, too dark, out of focus, etc.).

2.1 Creating a color profile

Our goal when creating a color profile is to have every image look similarly to the rest of the batch. When RawTherapee launches, it will bring us to the last batch we worked on by default. Be sure to navigate to the intended batch. Note when there is more than one folder within the batch. We must adjust every image in each folder before sending the batch into the pipeline.

It is good practice to scan through all images to get an idea of what they are looking like, if there are any outliers, or anything to make note of.

2.2 Look for a Color Checker

We want to use an image with a color checker, if one is available within the batch. Choose the most average looking picture (lighting and color-wise). If they all look reasonably similar, we can choose whichever we want.

*In Semifield applications, a smaller Color Checker lanyard is typically used. The Field Color Checker might be used instead, but the smaller one is preferred for this application. Either is acceptable.

2.12 Bring in a Recent profile & Color Picker

To bring in a recent profile, find the folder icon towards the top right of the screen, located to the left of the save button. This will open a window containing a list of saved profiles within the local user. Choose the most recent date in relation to the image being worked on.

Once a previous profile is loaded in and applied, use the color picker tool. The Color Picker tool is found on the tool bar above the working image or use the White Balance tab on the panel to the right. Choose “Pick”.

Only one square on the Color Checker is used for in this application. Referring to the image above, the gray residing two squares above the black is the color to use. Use this gray for every profile. Choose the hand icon to return to normal cursor mode.

2.13 Adjust exposure & saturation

Now, navigate into the Exposure tab. This is the only tab used in preprocessing, aside from the tab to access the Color Picker. There are only a few helpful settings to toggle here.

The goal is to create an image that looks most closely like a real-life view of these plants. The human eye is the most accurate check, but it can be helpful to reference the color graph located on the top left of the scree. It shifts when settings are changed. A good rule of thumb is for the peak of the three curves to land near this region of the graph.

*Laptop screens and Monitor settings will impact perception of images. Try to use the same screen when creating color profiles.

• Exposure Compensation & Highlight Compression

  • These two go hand-in-hand. Exposure compensation should always be set to slightly more than Highlight Compression

• Lightness

  • Has a similar effect to Exposure Compensation

• Saturation

  • Toggle keeping in mind the intensity of colors on the color checker.

3. Running the pipeline

When the color profile has been saved, we must send it through the pipeline. Be sure to be navigated to the Semifield folder and make a new screen, use uppercase r (-R). If returning to an already existing screen, use lowercase r (-r) with the appropriate screen name.

cd ..
cd psa_images/semifield_tools/
screen -R <batch_name>
python3 semifield_CLOUDY_pipeline.py <batch_name>

To exit the screen, use CTRL+A D

Check on a batch sent into the pipeline by using this string of commands:

cd psa_images/semifield_tools/
screen -r <batch_name>

If the pipeline has either been completed running or there has been an error running it, there will be an input line. If there is no input line, the process has not completed.

To exit the screen, use CTRL+A D

3.1 Checking the success of the pipeline

If there is an indication the pipeline is finished running, we want to crosscheck the # of images in the original batch with the # of JPGs and PP3s created.

We can use both of these lines respectively:

ls /mnt/research-projects/s/screberg/longterm_images2/semifield-developed-images/<batch>/images/*.jpg | wc -l

ls /mnt/research-projects/s/screberg/longterm_images2/semifield-developed-images/<batch>/images/*.pp3 | wc -l

And compare to the # of ARWs using this:

ls /mnt/research-projects/s/screberg/longterm_images2/semifield-upload/<batch>/*.ARW | wc -l

As long as the output # we get from each of these commands are equivalent, we know we have successfully applied the color profile to every image in the batch.

If the # of JPGs and PP3s are not equivalent to the original # of images, try running the pipeline again within the screen. Double check the name of the batch that was typed in, this is where errors are likely to happen. Also, be sure to be navigated to the Semifield folder. If we are within the Field folder, the pipeline will not be successful.

5. Update Backlog Spreadsheet

Change the progress column to the appropriate term to indicate where that batch is in the process. Make any notes/check box when done, if applicable.

When entire batch is fully completed, the row can be hidden by right clicking on the row #, then click hide.

4. Blob Analyzer (????)

4.1 Installing conda (miniconda) 

Use these commands below from the command line in the SUNNY server:

mkdir -p ~/miniconda3
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda3/miniconda.sh
bash ~/miniconda3/miniconda.sh -b -u -p ~/miniconda3
rm  ~/miniconda3/miniconda.sh
~/miniconda3/bin/conda init bash

Close the remote connection, then SSH back in

4.2 Setting up a conda environment

Still in the command line inside of SUNNY:

conda create -n <your env name> python=3.10
conda activate <your env name>
cd /home/psa_images/analyze_blobs
pip install -r requirements.txt

4.3 Running the code 

Your conda environment must be activated to run the code.

Run these in the command line:

conda activate <your env name>
cd /home/psa_images/analyze_blobs
python ANALYZE_BLOB.py

1. Creates 2 time stamped files in the “results” folder in the main project directory. 

   1. A text files that provides a general summary of missing batches

   2. A csv file that can be viewed in excel or google docs

5. Common & Useful Commands

Available batches:

ls /mnt/research-projects/s/screberg/longterm_images2/semifield-upload/ -lh

# of images in batch:

ls /mnt/research-projects/s/screberg/longterm_images2/semifield-upload/<batch>/*.ARW | wc -l

# of JPGs processed:

ls /mnt/research-projects/s/screberg/longterm_images2/semifield-developed-images/<batch>/images/*.jpg | wc -l

# of PP3 files created:

ls /mnt/research-projects/s/screberg/longterm_images2/semifield-developed-images/<batch>/images/*.pp3 | wc -l