2. How to Train a Segmentation Model¶
2.1. Installation¶
Install Google Deeplab
We use Google DeepLab to train our model.
Follow the installation instructions here .
tensorflow-gpu==1.12.0 and cuda version = 9.0 is recommended if an effor occurs while installing.
Download the dataset
We use Cityscapes Dataset for outdoor training data, and ADE20K Dataset for indoor training data. To download Cityscapes, register the website and download manualy or use
wget --keep-session-cookies --save-cookies=cookies.txt --post-data 'username=yourusername&password=yourpassword&submit=Login' https://www.cityscapes-dataset.com/login/
wget --load-cookies cookies.txt --content-disposition https://www.cityscapes-dataset.com/file-handling/?packageID=1,3
where the package ID needed are 1 and 3.
Follow the steps bellow to download ADE20K dataset.
Run the dataset
Steps to run Cityscapes can be found here .
Steps to run and download ADE20K can be found here .
We use initial checkpoint m`obilenetv2_coco_cityscapes_trainfine <http://download.tensorflow.org/models/deeplabv3_mnv2_cityscapes_train_2018_02_05.tar.gz>`_ provided by TensorFlow DeepLab Model Zoo .
Our parameters are
python deeplab/train.py \
--logtostderr \
--training_number_of_steps=90000 \
--train_split="train" \
--model_variant="mobilenet_v2" \
--output_stride=16 \
--train_crop_size=321 \
--train_crop_size=321 \
--train_batch_size=16 \
--fine_tune_batch_norm=true \
--dataset="cityscapes" \
--tf_initial_checkpoint=${PATH_TO_INITIAL_CHECKPOINT} \
--train_logdir=${PATH_TO_TRAIN_DIR} \
--dataset_dir=${PATH_TO_DATASET}
Train your own data
In addition to Cityscapes dataset, we also trained on images collected and annotated by students on our own campas. Add images to Cityscapes dataset in gtfine and leftImg8bit folders, and convert to TFRecord with convert_cityscapes.sh. Remember to convert annotation colors to trainIDs. A dictionary of trainIDs and anotation colors is shown below.
Label = namedtuple( 'Label' , ['name','id','trainId','category','categoryId','hasInstances','ignoreInEval','color',] )
labels = [
# name id trainId category catId hasInstances ignoreInEval color
Label( 'unlabeled' , 0 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ),
Label( 'ego vehicle' , 1 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ),
Label( 'rectification border' , 2 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ),
Label( 'out of roi' , 3 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ),
Label( 'static' , 4 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ),
Label( 'dynamic' , 5 , 255 , 'void' , 0 , False , True , (111, 74, 0) ),
Label( 'ground' , 6 , 255 , 'void' , 0 , False , True , ( 81, 0, 81) ),
Label( 'road' , 7 , 0 , 'flat' , 1 , False , False , (128, 64,128) ),
Label( 'sidewalk' , 8 , 1 , 'flat' , 1 , False , False , (244, 35,232) ),
Label( 'parking' , 9 , 255 , 'flat' , 1 , False , True , (250,170,160) ),
Label( 'rail track' , 10 , 255 , 'flat' , 1 , False , True , (230,150,140) ),
Label( 'building' , 11 , 2 , 'construction' , 2 , False , False , ( 70, 70, 70) ),
Label( 'wall' , 12 , 3 , 'construction' , 2 , False , False , (102,102,156) ),
Label( 'fence' , 13 , 4 , 'construction' , 2 , False , False , (190,153,153) ),
Label( 'guard rail' , 14 , 255 , 'construction' , 2 , False , True , (180,165,180) ),
Label( 'bridge' , 15 , 255 , 'construction' , 2 , False , True , (150,100,100) ),
Label( 'tunnel' , 16 , 255 , 'construction' , 2 , False , True , (150,120, 90) ),
Label( 'pole' , 17 , 5 , 'object' , 3 , False , False , (153,153,153) ),
Label( 'polegroup' , 18 , 255 , 'object' , 3 , False , True , (153,153,153) ),
Label( 'traffic light' , 19 , 6 , 'object' , 3 , False , False , (250,170, 30) ),
Label( 'traffic sign' , 20 , 7 , 'object' , 3 , False , False , (220,220, 0) ),
Label( 'vegetation' , 21 , 8 , 'nature' , 4 , False , False , (107,142, 35) ),
Label( 'terrain' , 22 , 9 , 'nature' , 4 , False , False , (152,251,152) ),
Label( 'sky' , 23 , 10 , 'sky' , 5 , False , False , ( 70,130,180) ),
Label( 'person' , 24 , 11 , 'human' , 6 , True , False , (220, 20, 60) ),
Label( 'rider' , 25 , 12 , 'human' , 6 , True , False , (255, 0, 0) ),
Label( 'car' , 26 , 13 , 'vehicle' , 7 , True , False , ( 0, 0,142) ),
Label( 'truck' , 27 , 14 , 'vehicle' , 7 , True , False , ( 0, 0, 70) ),
Label( 'bus' , 28 , 15 , 'vehicle' , 7 , True , False , ( 0, 60,100) ),
Label( 'caravan' , 29 , 255 , 'vehicle' , 7 , True , True , ( 0, 0, 90) ),
Label( 'trailer' , 30 , 255 , 'vehicle' , 7 , True , True , ( 0, 0,110) ),
Label( 'train' , 31 , 16 , 'vehicle' , 7 , True , False , ( 0, 80,100) ),
Label( 'motorcycle' , 32 , 17 , 'vehicle' , 7 , True , False , ( 0, 0,230) ),
Label( 'bicycle' , 33 , 18 , 'vehicle' , 7 , True , False , (119, 11, 32) ),
Label( 'license plate' , -1 , -1 , 'vehicle' , 7 , False , True , ( 0, 0,142) ),
]
Using your trained model
After exporting your model, tar frozen_inference_graph.pb with
tar -zcvf mymodel.tar.gz frozen_inference_graph.pb
An inference example can be found here <https://colab.research.google.com/github/tensorflow/models/blob/master/research/deeplab/deeplab_demo.ipynb> _.