sanghoon / tracking_via_colorization

Tracking via Colorization Tensorflow implementatin of Tracking Emerges by Colorizing Videos

Tracking via Colorization

• colorization sample

• tracking segmentation sample

Introduction

This TensorFlow implementation is designed with these goals:

• Tracking via Colorization Tensorflow implementatin of Tracking Emerges by Colorizing Videos

How to Use

Clustering

python3 bin/clustering.py -k 16 -n 10000 -o datas/centroids/centroids_16k_cifar10_10000samples.npy

Train

• colorizer

python3 bin/train_colorizer.py --model-dir models/colorizer
tensorboard --host 0.0.0.0 --port 6006 --logdir models

• cifar10

python3 bin/train_estimator_cifar10.py --model-dir models/test

Predict

• colorizer

python3 bin/test_colorizer.py --checkpoint models/test/model.ckpt-100000 --scale 1 --name davis -o results/davis/

Prerequisite

Should install below libraries.

• Tensorflow >= 1.10
• opencv >= 3.0

And install below dependencies.

apt install -y libsm6 libxext-dev libxrender-dev libcap-dev
apt install -y ffmpeg
pip install -r requirements.txt