|
--- |
|
license: wtfpl |
|
language: |
|
- en |
|
- ur |
|
tags: |
|
- facial |
|
- recognization |
|
- face |
|
- ML |
|
- Ai |
|
--- |
|
Facial Recognition Model |
|
|
|
Model Information |
|
|
|
Name: Facial Means |
|
|
|
Type: Convolutional Neural Network (CNN) |
|
|
|
Framework: TensorFlow |
|
|
|
Dataset: Celebrity Faces Dataset |
|
|
|
Dataset Path: /content/drive/MyDrive/beard_dataset/celb_dataset/ |
|
|
|
Model Save Path: /content/drive/MyDrive/beard_dataset/celebrity_model.h5 |
|
|
|
Image Dimensions: 224 x 224 pixels |
|
|
|
Batch Size: 32 |
|
Data Augmentation |
|
Rescale: 1./255 |
|
Shear Range: 0.2 |
|
Zoom Range: 0.2 |
|
Horizontal Flip: True |
|
Validation Split: 20% |
|
Model Architecture |
|
Layer (type) |
|
Output Shape Param # |
|
=============================================================== |
|
conv2d (Conv2D) (None, 222, 222, 32) 896 |
|
max_pooling2d (MaxPooling2D) (None, 111, 111, 32) 0 |
|
conv2d_1 (Conv2D) (None, 109, 109, 64) 18496 |
|
max_pooling2d_1 (MaxPooling2D)(None, 54, 54, 64) 0 |
|
conv2d_2 (Conv2D) (None, 52, 52, 128) 73856 |
|
max_pooling2d_2 (MaxPooling2D)(None, 26, 26, 128) 0 |
|
flatten (Flatten) (None, 86528) 0 |
|
dense (Dense) (None, 128) 11075712 |
|
dense_1 (Dense) (None, 6) 774 |
|
=============================================================== |
|
|
|
Total params: 11,170,734 |
|
Trainable params: 11,170,734 |
|
Non-trainable params: 0 |
|
Model Compilation |
|
Optimizer: Adam |
|
Loss Function: Categorical Crossentropy |
|
Metrics: Accuracy |
|
|
|
Training |
|
|
|
Epochs: 10 |
|
|
|
Steps per Epoch: Calculated based on the training dataset size and batch size. |
|
Validation Steps: Calculated based on the validation dataset size and batch size. |
|
Model Save |
|
The trained model is saved at /content/drive/MyDrive/beard_dataset/celebrity_model.h5. |
|
Conclusion |
|
The facial recognition model has been trained on a celebrity faces dataset using TensorFlow. The model architecture includes convolutional and pooling layers, followed by fully connected layers for classification. The training process involves data augmentation and achieves satisfactory accuracy. |
|
|
|
Model training code provided by Muhammad Sajjad Rasool. |
