shamangary / Keras Morph2 Age Estimation
Keras implementation for MORPH2 dataset
Stars: ✭ 17
Programming Languages
python
139335 projects - #7 most used programming language
Projects that are alternatives of or similar to Keras Morph2 Age Estimation
Fsa Net
[CVPR19] FSA-Net: Learning Fine-Grained Structure Aggregation for Head Pose Estimation from a Single Image
Stars: ✭ 469 (+2658.82%)
Mutual labels: face
Extreme 3d faces
Extreme 3D Face Reconstruction: Looking Past Occlusions
Stars: ✭ 653 (+3741.18%)
Mutual labels: face
Emopy
A deep neural net toolkit for emotion analysis via Facial Expression Recognition (FER)
Stars: ✭ 744 (+4276.47%)
Mutual labels: face
Ringnet
Learning to Regress 3D Face Shape and Expression from an Image without 3D Supervision
Stars: ✭ 504 (+2864.71%)
Mutual labels: face
Pico
A minimalistic framework for real-time object detection (with a pre-trained face detector)
Stars: ✭ 561 (+3200%)
Mutual labels: face
React Native Fingerprint Scanner
Provide Fingerprint, Touch ID, and Face ID Scanner for React Native (Compatible with both Android and iOS)
Stars: ✭ 704 (+4041.18%)
Mutual labels: face
Face Pose Net
Estimate 3D face pose (6DoF) or 11 parameters of 3x4 projection matrix by a Convolutional Neural Network
Stars: ✭ 464 (+2629.41%)
Mutual labels: face
Uiimageview Betterface
a UIImageView category to let the picture-cutting with faces showing better
Stars: ✭ 790 (+4547.06%)
Mutual labels: face
Faceswap
Real-time FaceSwap application built with OpenCV and dlib
Stars: ✭ 611 (+3494.12%)
Mutual labels: face
Awesome Face
😎 face releated algorithm, dataset and paper
Stars: ✭ 739 (+4247.06%)
Mutual labels: face
Face segmentation
Deep face segmentation in extremely hard conditions
Stars: ✭ 510 (+2900%)
Mutual labels: face
R2cnn faster Rcnn tensorflow
Rotational region detection based on Faster-RCNN.
Stars: ✭ 548 (+3123.53%)
Mutual labels: face
Tensorflow Face Detection
A mobilenet SSD based face detector, powered by tensorflow object detection api, trained by WIDERFACE dataset.
Stars: ✭ 711 (+4082.35%)
Mutual labels: face
Facecropper
✂️ Crop faces, inside of your image, with iOS 11 Vision api.
Stars: ✭ 479 (+2717.65%)
Mutual labels: face
Biometricauthentication
Use Apple FaceID or TouchID authentication in your app using BiometricAuthentication.
Stars: ✭ 746 (+4288.24%)
Mutual labels: face
Vrn
👨 Code for "Large Pose 3D Face Reconstruction from a Single Image via Direct Volumetric CNN Regression"
Stars: ✭ 4,391 (+25729.41%)
Mutual labels: face
Jeelizweboji
JavaScript/WebGL real-time face tracking and expression detection library. Build your own emoticons animated in real time in the browser! SVG and THREE.js integration demos are provided.
Stars: ✭ 835 (+4811.76%)
Mutual labels: face
Human Detection And Tracking
Human-detection-and-Tracking
Stars: ✭ 753 (+4329.41%)
Mutual labels: face
Keras-MORPH2-age-estimation
Keras implementation for MORPH2 dataset age estimation.
This project contains Mobilenet and Densenet with regression and DEX framework.
Update (2017/12/1)
- Fix inconsistent label problem.
- Add face align in preprocessing.
Update(2017/11/27)
- Change training default epoch to 90
- Decay learning rate at epoch [30,60]
How to run?
-
Step.1 Download landmarks from https://github.com/xyfeng/average_portrait Unzip it and move to './landmarks'
-
Step.2 Download MORPH2 dataset https://www.faceaginggroup.com/morph/ Unzip it under './morph2'
You have to apply for the dataset. No easy way to download it unfortunately :(
- Step.3 Preprocess the dataset (change isPlot inside TYY_MORPH_create_db.py to True if you want to see the process)
python TYY_MORPH_create_db.py --output morph_db.npz
- Step.4 Run the training and evalutation (change netType inside TYY_train_MORPH.py for different networks)
KERAS_BACKEND=tensorflow python TYY_train_MORPH.py --input ./morph_db.npz
Training and evaluation
-
Training ratio: 0.8
-
Validation ratio: 0.2
-
Evaluation metric: Mean-absoluate-error (MAE) -> name: val_pred_a_mean_absolute_error
-
Output example:
pred_a_softmax_loss: 2.4073 - pred_a_loss: 9.4221 - pred_a_softmax_acc: 0.1183 - pred_a_mean_absolute_error: 9.4221 - val_loss: 2.4423 - val_pred_a_softmax_loss: 2.4423 - val_pred_a_loss: 9.4864 - val_pred_a_softmax_acc: 0.1339 - val_pred_a_mean_absolute_error: 9.4864
Parameters
- DEX: num_neu is the output dimension of the classfication training part. Range of num_neu: [1~101]
- Mobilenet: alpha is the paramters to control the network size. Recommended value of alpha: 1, 0.5, 0.25
- Densenet: densenet_depth is the depth of the network (Obviously~~)
Dependencies
- Keras
- Tensorflow
- anaconda
- python3
- opencv3
- dlib
- moviepy
- pytables
References
- https://github.com/yu4u/age-gender-estimation
- https://github.com/titu1994/DenseNet
- R. Rothe, R. Timofte, and L. V. Gool, "Deep expectation of real and apparent age from a single image without facial landmarks," IJCV, 2016.
- https://github.com/fchollet/keras/blob/master/keras/applications/mobilenet.py
- https://github.com/xyfeng/average_portrait
Note that the project description data, including the texts, logos, images, and/or trademarks,
for each open source project belongs to its rightful owner.
If you wish to add or remove any projects, please contact us at [email protected].