MCCCS docker run on example data

Using docker to perform the MCCCS example data analysis pipeline

An introduction to the example use-cases can be found on the MCCCS website: http://openimageanalysisgroup.github.io/MCCCS/subsites/exp.html

Download of the image and start of the docker container

To interact with the docker command line, start the Terminal application on Mac or Linux, on Windows start the Windows PowerShell. Then type the following commands.

1. Download image

   docker pull klukas/mcccs

2. Create, start and exit container

   docker run -it --name mcccs klukas/mcccs /bin/bash

   exit

   If you are working behind a proxy use the following options (additionally dns server specifications are possible) to start the docker container:

   docker run -e "http_proxy=http://xyz.xyz.xyz:1234" -e "https_proxy=http://xyz.xyz.xyz:1234" --dns www.xxx.yyy.zzz -it --name mcccs klukas/mcccs /bin/bash

3. Check list of Containers

   docker ps -a

4. Start and connect to existing stopped container

   docker start -ai mcccs

Download example data and libraries

Within the MCCCS docker bash shell (use step 4 from previous section to start the shell) execute the following commands.

1. cd start
2. ./prepare_datasets.sh

Hyper-spectral example analysis 1 (classification)

Within the MCCCS docker bash shell (still open from previous section, or from step 4 of section 1) perform the following operations.

1. cd /start/example_data/hyper_example_1_classification/
2. bash process_hyper_example_1_classification.sh

**The analysis script will finish after about 10 minutes. The result image file is './stack_images/dc/classified.png'. Numeric analysis results are saved as './stack_images/dc/all_results.csv' **

Use the following docker commands (from another Terminal or PowerShell window) to extract it to your operating systems file system:

docker cp mcccs:/start/example_data/hyper_example_1_classification/stack_images/dc/classified.png .

docker cp mcccs:/start/example_data/hyper_example_1_classification/stack_images/all_results.csv .

Hyper-spectral example analysis 2 (clustering)

Within the MCCCS docker bash shell (still open from previous section, or from step 4 of section 1) perform the following operations.

1. cd /start/example_data/hyper_example_2_clustering/
2. bash process_hyper_example_2_clustering.sh

The analysis script will finish after about 2 or 3 minutes. The result image file is './stack_images/dc/classified.png'.

Use the following docker command (from another Terminal or PowerShell window) to extract it to your operating systems file system:

docker cp mcccs:/start/example_data/hyper_example_2_clustering/stack_images/dc/classified.png .

This is one command-line call, if displayed on several lines, enter the command without line break.

Foreground/Background Separation example

[Operation time for this example is currently about 5 hours]

Within this example three data sets are processed. For a number of training images the foreground has been annotated. Based on this training data a classifier is trained (using Weka) and predictions are calculated for the test data.

Use the MCCCS docker bash shell (still open from previous section, or from step 4 of section 1) to perform the following operations. Once the script is running you may observe the progress and operation status using the commands explained in the next section of this page.

1. cd /start/example_data/segmentation_example_1_classification/
2. bash process_process_segmentation_example_1_classification.sh

Investigate current operation status of the running docker container

You may investigate the running processes (e.g. operation status of the MCCCS scripts), by connecting a second time to the container and then by starting e.g. top:

1. docker exec -it mcccs /bin/bash
2. Within the bash terminal start top

Stop container

docker stop mcccs

Example session on Windows 10 (screenshot)