Programming

:Project Description

Specifically, the task focuses on binary classification tasks using a provided dataset, which is a subset of the well-known CelebA dataset. The dataset contains 5,000 celebrity images along with annotations for 40 facial attributes, including gender and emotion (smiling or not smiling).

To accomplish the binary classification tasks, the project utilizes Python, Scikit-Learn, and dlib. The dlib library's face detector is employed to extract facial landmarks, which serve as features for the classification models. Scikit-Learn is used to train binary classifiers based on the extracted features.

:My Contribution

Throughout the project, I conducted a thorough analysis of the CelebA dataset, ensuring its suitability for the classification tasks. I performed data preprocessing to address any missing or noisy data, and I split the dataset into training and validation sets for model evaluation.

Using the dlib library's face detector, I successfully detected and located faces in the images. I employed the shape predictor to extract facial landmarks as features, which proved crucial for the classification tasks. I visualized the detected faces and landmarks to verify their accuracy and gain insights into the data.

For gender detection, I designed and implemented a binary classification model using Scikit-Learn. The model was trained on the extracted facial landmarks, and I thoroughly evaluated its performance. This evaluation involved analyzing training errors, validation errors, and hyperparameter tuning. I also experimented with multiple models for gender detection, comparing their performance and providing justifications for the chosen methodology.

Similarly, for emotion detection (smiling or not smiling), I developed another binary classification model using Scikit-Learn and the extracted facial landmarks. I trained the model and assessed its performance using appropriate evaluation metrics.

In conclusion, leveraging my programming skills and understanding of ML systems, I successfully completed the data analysis, preprocessing, face detection, feature extraction, and classification tasks for both gender and emotion detection.