Skip to content

Latest commit

 

History

History
102 lines (77 loc) · 3.33 KB

README.md

File metadata and controls

102 lines (77 loc) · 3.33 KB
Raw

Age Estimation

This sub-project focuses on improving the accuracy of age estimation.

Dataset Preparation

APPA-REAL Dataset

Follow the instructions here to download and extract the dataset.

UTK Face Dataset

Firstly, download images from the website of the UTKFace dataset. part1.tar.gz, part2.tar.gz, and part3.tar.gz can be downloaded from In-the-wild Faces in Datasets section. Then, extract the archives:

tar zxf part1.tar.gz
tar zxf part2.tar.gz
tar zxf part3.tar.gz

Finally, run the following script to create the training data:

python3 utkface/create_db_utkface_with_margin.py --input [PATH_TO_DATASET_DIR] --output [OUTPUT_DIR]

[PATH_TO_DATASET_DIR] should be a directory that includes part1, part2, and part3 directories. The cropped face images with margin will be created in [OUTPUT_DIR].

Training

python3 train.py --appa_dir [PATH_to_appa-real-release] --utk_dir [PATH_TO_UTK_CROPPED_FACE_DIR] --nb_epochs 100

Options:

usage: train.py [-h] --appa_dir APPA_DIR [--utk_dir UTK_DIR]
                [--output_dir OUTPUT_DIR] [--batch_size BATCH_SIZE]
                [--nb_epochs NB_EPOCHS] [--lr LR] [--opt OPT]
                [--model_name MODEL_NAME]

This script trains the CNN model for age estimation.

optional arguments:
  -h, --help            show this help message and exit
  --appa_dir APPA_DIR   path to the APPA-REAL dataset (default: None)
  --utk_dir UTK_DIR     path to the UTK face dataset (default: None)
  --output_dir OUTPUT_DIR
                        checkpoint dir (default: checkpoints)
  --batch_size BATCH_SIZE
                        batch size (default: 32)
  --nb_epochs NB_EPOCHS
                        number of epochs (default: 30)
  --lr LR               learning rate (default: 0.1)
  --opt OPT             optimizer name; 'sgd' or 'adam' (default: sgd)
  --model_name MODEL_NAME
                        model name: 'ResNet50' or 'InceptionResNetV2'
                        (default: ResNet50)

Result

Currently the best MAE (against apparent age) is 4.410. This model can be trained by:

python3 train.py --appa_dir APPA_DIR --opt adam --lr 0.001 --nb_epochs 100

weights: https://github.com/yu4u/age-gender-estimation/releases/download/v0.5/age_only_resnet50_weights.061-3.300-4.410.hdf5

Demo

Run the demo script (requires web cam). You can use --image_dir [IMAGE_DIR] option to use images in the [IMAGE_DIR] directory instead.

python3 demo.py

The pretrained model for TensorFlow backend will be automatically downloaded to the pretrained_models directory.

optional arguments:
  -h, --help            show this help message and exit
  --model_name MODEL_NAME
                        model name: 'ResNet50' or 'InceptionResNetV2'
                        (default: ResNet50)
  --weight_file WEIGHT_FILE
                        path to weight file (e.g.
                        age_only_weights.029-4.027-5.250.hdf5) (default: None)
  --margin MARGIN       margin around detected face for age-gender estimation
                        (default: 0.4)
  --image_dir IMAGE_DIR
                        target image directory; if set, images in image_dir
                        are used instead of webcam (default: None)