Skip to content

Latest commit

 

History

History
45 lines (38 loc) · 2.61 KB

README.md

File metadata and controls

45 lines (38 loc) · 2.61 KB

TensorFlow 1.1.0 Python 3.5 Jupyter Notebook

Windows x64 cuDNN v5 CUDA 8.0

Human Gender Prediction

Chonnam National University
2017 AI Class
Professor Lee Chilwoo
Student: Nguyen Hai Duong
Target: predict human gender using Convolutional Neural Network

Gender prediction examples

alt text

Confusion matrix on Wiki testing set

alt text

How to run source code

For a complete training (at least 8GB memory is required)

  1. Download the Jupyter notebook gender_prediction.ipynb and save it to a specific path (called GD_PATH)
  2. In GD_PATH, create folder gender/wiki_crop
  3. Download the processed WIKI dataset [1] and save them to GD_PATH/gender/wiki_crop
  4. Download additional testing images and store them in GD_PATH
  5. Open gender_prediction.ipynb with Jupyter and run all cells

For testing using trained model

  1. Download the Jupyter notebook gender_prediction_testing.ipynb and save it to a specific path (called GD_PATH)
  2. Download trained model on WIKI dataset [1] and save it to GD_PATH
  3. In GD_PATH, create folder gender/wiki_crop
  4. Download WIKI testingset [1] including 64_64_11938_4098_testing_x_onehot.npy, and 64_64_11938_4098_testing_y_onehot.npy and store them in GD_PATH/gender/wiki_crop
  5. Download additional testing images and store them in GD_PATH
  6. Open gender_prediction_testing.ipynb with Jupyter and run all cells

References

[1] Rasmus Rothe, Radu Timofte, and Luc Van Gool, "Deep expectation of real and apparent age from a single image without facial landmarks," International Journal of Computer Vision (2016)

Personal information

Nguyen Hai Duong
Supervisor: Professor Kim Soo-Hyung
Pattern Recognition Lab
Chonnam National University, Korea
E-mail: nhduong_3010@live.com