This project focuses on classifying the gender of individuals from facial images. It employs a combination of techniques including transfer learning, fine-tuning, and custom CNN models.
-
Transfer Learning with VGG16
- Trained on a dataset of 12,000 face images.
- Achieved 55% accuracy, but was slower than desired.
-
Fine-tuned Model from Hugging Face (rizvandwiki)
- Improved performance using a pre-trained model specifically designed for gender detection.
- Provided a good balance of accuracy and speed.
-
Face Detection with OpenCV
- Utilized OpenCV's in-built harcascade model for face detection.
After training multiple CNN models on a dataset of 12,000+ images with corresponding testing on 5,000+ samples, the following results were obtained:
-
Model 1:
- Train Accuracy: 86%
- Test Accuracy: 85%
-
Model 2:
- Train Accuracy: 90%
- Test Accuracy: 89%
-
Model 3:
- Train Accuracy: 85%
- Test Accuracy: 83%
Given these results, the second model was selected for the final implementation. It demonstrated a commendable balance between training and testing accuracy, achieving 90% and 89% respectively.
| Model | Train Accuracy | Test Accuracy |
|---|---|---|
| Fine-tuned VGG16 | 61% | 55% |
| rizvandwiki's Model | - | - |
| Custom CNN Model 1 | 86% | 85% |
| Custom CNN Model 2 | 90% | 89% |
| Custom CNN Model 3 | 85% | 83% |
- Python 3.x
- OpenCV
- Tensorflow >=2.4
- Transformers
- ThreadPoolExecutor
- keras.applications
- download the model & keep it in your working Directory
- Gender Detection Model by rizvandwiki
from transformers import AutoFeatureExtractor, AutoModelForImageClassification
extractor = AutoFeatureExtractor.from_pretrained("rizvandwiki/gender-classification-2")
model = AutoModelForImageClassification.from_pretrained("rizvandwiki/gender-classification-2")
inputs = extractor(images = train_images[image_name], return_tensors="pt" )
with torch.no_grad():
logits = model(**inputs).logits
predicted_label = logits.argmax(-1).item()
label = model.config.id2label[predicted_label]
- Clone the Repository
git clone https://github.com/your_username/your_project.git cd your_project - Install Dependencies
pip install -r requirements.txt
- Run the main file
python gender_detection_app.py
This Project was a part of a College Assignment (Partial). For any inquiry please feel free to concat me or you can raise an issue in the GitHub issue section.