Counting 1,868 Big Data & Machine Learning Frameworks, Toolsets, and Examples...
Suggestion? Feedback? Tweet @stkim1

Last Commit
Oct. 15, 2017
Jul. 29, 2017


TensorFlow implementation of ENet ( based on the official Torch implementation ( and the Keras implementation by PavlosMelissinos (, trained on the Cityscapes dataset (

  • Youtube video of results (

  • demo video with results

  • The results in the video can obviously be improved, but because of limited computing resources (personally funded Azure VM) I did not perform any further hyperparameter tuning.

You might get the error "No gradient defined for operation 'MaxPoolWithArgmax_1' (op type: MaxPoolWithArgmax)". To fix this, I had to add the following code to the file /usr/local/lib/python2.7/dist-packages/tensorflow/python/ops/

def _MaxPoolGradWithArgmax(op, grad, unused_argmax_grad):  
  return gen_nn_ops._max_pool_grad_with_argmax(op.inputs[0], grad, op.outputs[1], op.get_attr("ksize"), op.get_attr("strides"), padding=op.get_attr("padding"))  


  • ASSUMES: that all Cityscapes training (validation) image directories have been placed in data_dir/cityscapes/leftImg8bit/train (data_dir/cityscapes/leftImg8bit/val) and that all corresponding ground truth directories have been placed in data_dir/cityscapes/gtFine/train (data_dir/cityscapes/gtFine/val).
  • DOES: script for performing all necessary preprocessing of images and labels.

  • ASSUMES: that has already been run.
  • DOES: contains the ENet_model class.

  • ASSUMES: -
  • DOES: contains a number of functions used in different parts of the project.

  • ASSUMES: that has already been run.
  • DOES: script for training the model.

  • ASSUMES: that has already been run.
  • DOES: runs a model checkpoint (set in line 56) on all frames in a Cityscapes demo sequence directory (set in line 30) and creates a video of the result.

Training details:

  • In the paper the authors suggest that you first pretrain the encoder to categorize downsampled regions of the input images, I did however train the entire network from scratch.

  • Batch size: 4.

  • For all other hyperparameters I used the same values as in the paper.

  • Training loss:

  • training loss

  • Validation loss:

  • validation loss

  • The results in the video above was obtained with the model at epoch 23, for which a checkpoint is included in segmentation/training_logs/best_model in the repo.

Training on Microsoft Azure:

To train the model, I used an NC6 virtual machine on Microsoft Azure. Below I have listed what I needed to do in order to get started, and some things I found useful. For reference, my username was 'fregu856':



NV_GPU="$GPUIDS" nvidia-docker run -it --rm \
        -p 5584:5584 \
        --name "$NAME""$GPUIDS" \
        -v /home/fregu856:/root/ \
        tensorflow/tensorflow:latest-gpu bash
  • /root/ will now be mapped to /home/fregu856 (i.e., $ cd -- takes you to the regular home folder).

  • To start the image:

    • $ sudo sh
  • To commit changes to the image:

    • Open a new terminal window.
    • $ sudo docker commit fregu856_GPU0 tensorflow/tensorflow:latest-gpu
  • To stop the image when it’s running:

    • $ sudo docker stop fregu856_GPU0
  • To exit the image without killing running code:

    • Ctrl-P + Q
  • To get back into a running image:

    • $ sudo docker attach fregu856_GPU0
  • To open more than one terminal window at the same time:

    • $ sudo docker exec -it fregu856_GPU0 bash
  • To install the needed software inside the docker image:

    • $ apt-get update
    • $ apt-get install nano
    • $ apt-get install sudo
    • $ apt-get install wget
    • $ sudo apt-get install libopencv-dev python-opencv
    • Commit changes to the image (otherwise, the installed packages will be removed at exit!)