Skip to content

MMSegmentation

MMSegmentation is a comprehensive and flexible framework for image segmentation that offers a wide variety of architectures. Custom docker images with additional tools are available from here:

https://github.com/waikato-datamining/mmsegmentation

Prerequisites#

Make sure you have the directory structure created as outlined in the Prerequisites.

Data#

In this example, we will use the CamVid-12 dataset, which consists of still images from dashcam videos from a city environment (12 different labels).

Download the dataset from the following URL into the data directory and extract it:

https://datasets.cms.waikato.ac.nz/ufdl/data/camvid12/camvid12-grayscale.zip

Once extracted, rename the grayscale directory to camvid-grayscale.

Now we have to convert the format from grayscale into indexed PNG, which MMSegmentation uses. We can do this by using the image-dataset-converter library. At the same time, we can split the dataset into train, validation and test subsets.

From within the applied_deep_learning directory, run the following command:

docker run --rm -u $(id -u):$(id -g) \
  -v `pwd`:/workspace \
  -t waikatodatamining/image-dataset-converter:0.0.4 \
  idc-convert \
    -l INFO \
    from-grayscale-is \
      -i "/workspace/data/camvid-grayscale/*.png" \
      --labels sky building pole road pavement tree signsymbol fence car pedestrian bicyclist unlabelled \
    to-indexed-png-is \
      -o /workspace/data/camvid-indexed-split \
      --split_names train val test \
      --split_ratios 70 15 15

Training#

For training, we will use the following docker image:

waikatodatamining/mmsegmentation:0.30.0_cuda11.1

Training and inference are possible on just a CPU as well (though much, much slower). For utilizing a CPU you can use the following docker image:

waikatodatamining/mmsegmentation:0.30.0_cpu

The training script is called mmseg_train, for which we can invoke the help screen as follows:

docker run --rm -t waikatodatamining/mmsegmentation:0.30.0_cuda11.1 mmseg_train --help

It is good practice creating a separate sub-directory for each training run, with a directory name that hints at what dataset and model were used. So for our first training run, which will use mainly default parameters, we will create the following directory in the output folder:

camvid12-mmseg-pspnet50

Before we can train, we will need to obtain and customize a config file. Within the container, you can find example configurations for various architectures in the following directory:

/mmsegmentation/configs

Using the mmseg_config command, we can expand and dump one of these configurations for our own purposes:

docker run --rm \
  -u $(id -u):$(id -g) \
  --gpus=all \
  -v `pwd`:/workspace \
  -v `pwd`/cache:/.cache \
  -v `pwd`/cache/torch:/.cache/torch \
  -t waikatodatamining/mmsegmentation:0.30.0_cuda11.1 \
  mmseg_config \
  --config /mmsegmentation/configs/pspnet/pspnet_r50-d8_512x1024_40k_cityscapes.py \
  --output_config /workspace/output/camvid12-mmseg-pspnet50/pspnet_r50.py

Open the pspnet_r50.py file in a text editor and perform the following operations:

  • change num_classes to 12 (background not counted).
  • change dataset_type to ExternalDataset and any occurrences of type in the train, test, val sections of the data dictionary.
  • change data_root occurrences to /workspace/data/camvid-indexed-split (the directory above the train and val directories).
  • change img_dir occurrences to img_dir=data_root+'/DIR', with DIR being the appropriate train, val or test
  • do the same with ann_dir

Kick off the training with the following command:

docker run --rm \
  -u $(id -u):$(id -g) \
  --shm-size=8g \
  --gpus=all \
  -v `pwd`:/workspace \
  -v `pwd`/cache:/.cache \
  -v `pwd`/cache/torch:/.cache/torch \
  -e MMSEG_CLASSES="sky,building,pole,road,pavement,tree,signsymbol,fence,car,pedestrian,bicyclist,unlabelled" \
  -t waikatodatamining/mmsegmentation:0.30.0_cuda11.1 \
  mmseg_train \
  /workspace/output/camvid12-mmseg-pspnet50/pspnet_r50.py \
  --work-dir /workspace/output/camvid12-mmseg-pspnet50/runs

Predicting#

Using the mmseg_predict_poll script, we can batch-process images placed in the predictions/in directory as follows (e.g., from our test subset): 

docker run --rm \
  -u $(id -u):$(id -g) \
  --shm-size 8G \
  --gpus=all \
  -v `pwd`:/workspace \
  -v `pwd`/cache:/.cache \
  -v `pwd`/cache/torch:/.cache/torch \
  -e MMSEG_CLASSES="sky,building,pole,road,pavement,tree,signsymbol,fence,car,pedestrian,bicyclist,unlabelled" \
  -t waikatodatamining/mmsegmentation:0.30.0_cuda11.1 \
  mmseg_predict_poll \
  --model /workspace/output/camvid12-mmseg-pspnet50/runs/latest.pth \
  --config /workspace/output/camvid12-mmseg-pspnet50/pspnet_r50.py \
  --prediction_in /workspace/predictions/in \
  --prediction_out /workspace/predictions/out

Notes

  • The predictions can either be output in grayscale (default) or bluechannel format (--prediction_format).

  • You can view the predictions with the ADAMS Preview browser:

Example prediction

Screenshot

Screenshot