Skip to content

Latest commit

 

History

History
89 lines (65 loc) · 4.61 KB

README.md

File metadata and controls

89 lines (65 loc) · 4.61 KB
Raw

Unsupervised Learning of Object Landmarks through Conditional Image Generation

Tomas Jakab*, Ankush Gupta*, Hakan Bilen, Andrea Vedaldi (* equal contribution). Advances in Neural Information Processing Systems (NeurIPS) 2018.

Software that learns to discover object landmarks without any manual annotations. It automatically learns from images or videos and works across different datasets of faces, humans, and 3D objects.

Unsupervised Landmarks

Requirements

  • Linux
  • Python 2.7
  • TensorFlow 1.10.0. Other versions (1.*.*) are also likely to work
  • Torch 0.4.1
  • CUDA cuDNN. CPU mode may work but is untested
  • Python dependecies listed in requirements.txt

Getting Started

Installation

Clone this repository

git clone https://github.com/tomasjakab/imm && cd imm

Install Python dependecies by running

pip install --upgrade -r requirements.txt

Add the path to this codebase to PYTHONPATH

export PYTHONPATH=$PYTHONPATH:$(pwd)

Visualize Unsupervised Landmarks

Download trained models [0.9G] and set the path to them in configs/paths/default.yaml, option logdir.

Use Jupyter notebook examples/visualize.ipynb to run a model trained on AFLW dataset of faces that predicts 10 unsupervised landmarks.

Test Trained Models

We provide pre-trained models to re-produce the experimental results on facial landmark detection datasets (CelebA, MAFL, and AFLW). Please download them first as described in Getting Started/Visualize Unsupervised Landmarks.

CelebA and MAFL Datasets

Download CelebA [7.8G] dataset and set the path to it configs/paths/default.yaml, option celeba_data_dir. MAFL dataset is already included in CelebA download.

To test on MAFL dataset run

bash examples/test_mafl.sh <N>

This loads a model that was trained on CelebA dataset to predict N unsupervised landmarks (N can be set to 10, 30 or 50). It then trains a linear regressor from unsupervised landmarks to 5 labeled landmarks using MAFL training set and evaluates it on MAFL test set.

AFLW Dataset

Download AFLW [1.1G] dataset and set the path to it configs/paths/default.yaml, option aflw_data_dir.

To test on AFLW dataset run

bash examples/test_aflw.sh <N>

This loads a model that was trained on CelebA dataset and finetuned on AFLW dataset to predict N unsupervised landmarks (N can be set to 10, 30, or 50). It then trains a linear regressor from unsupervised landmarks to 5 labeled landmarks using AFLW training set and evaluates it on AFLW test set.

Training

If you wish to train your own model, please download VGG16 model [0.6G] that was pre-trained on colorization task and is needed for perceptual loss. This model comes from the paper Colorization as a Proxy Task for Visual Understanding, Larsson, Maire, Shakhnarovich, CVPR 2017. Set the path to this model in configs/paths/default.yaml, option vgg16_path. Also download and update the paths to the datasets as described above.

Set the option logdir in configs/paths/default.yaml to the location where you wish to store training logs and checkpoints.

CelebA Dataset

To train a model for N (e.g., N can be 10, 30 or anything else) unsupervised landmarks on CelebA dataset run

bash examples/train_celeba.sh <N>

AFLW Dataset

We first train on CelebA as described above, and then fine-tune on AFLW due to its small size.

To finetune a model for N unsupervised landmarks on AFLW dataset run

bash examples/train_aflw.sh <N> <celeba_checkpoint>

where celeba_checkpoint is the path to the model checkpoint trained on CelebA. This could be for example data/logs/celeba-10pts/model.ckpt.

Legacy Training and Evaluation Code

Test errors reported in the paper were obtained with a data pipline that was using MATLAB for image pre-processing. This codebase uses a Python re-implementation. Due to numerical differences, the test errors may slightly differ. If you wish to reproduce the exact numbers from the paper contact us at [email protected] to get this data pipeline (requires MATLAB).