KnowYourAge – DEX-based Age Estimation

This is a PyTorch implementation of an age prediction model based on the DEX (Deep EXpectation) approach using a VGG-16 backbone. The model was trained on the UTKFace dataset and predicts the expected age from a facial image.

🧠 Model Architecture

• Backbone: VGG-16 (pretrained on ImageNet)
• Output: 101-bin softmax, ages 0–100
• Final Prediction: Expected age = softmax-weighted sum of bins
• Loss Function: MSELoss
• Input Size: 224×224 RGB image
• Metric: Mean Absolute Error (MAE)

🗂️ Files

• pytorch_model.bin: Trained model weights
• config.json: Model metadata
• model.py: The PyTorch model class (optional but included here)
• README.md: You’re reading it!

📦 Usage

from model import DEXAgeModel
import torch

model = DEXAgeModel(num_bins=101)
state_dict = torch.load("pytorch_model.bin", map_location="cpu")
model.load_state_dict(state_dict)
model.eval()