Muhammad Ali Farooq

We propose SynAdult, a multimodal synthetic data generation framework designed to address the scarcity of diverse and privacy-compliant senior adult face datasets for biometric applications and facial analysis. Our pipeline begins with the rendering of high-fidelity 2D adult facial images using parameter-efficient LoRA-based tuning of the state-of-the-art hyperrealism Stable Diffusion XL (SDXL) model, producing photorealistic outputs across diverse ethnicities and age-specific features. Next, we integrate a video retargeting pipeline to synthesize temporally consistent head pose and facial expression sequences, ensuring naturalistic dynamics crucial for downstream video-based facial analysis. In the third stage, we generate neuromorphic event data to introduce a privacy-preserving modality aligned with real-world edge deployment scenarios, such as ambient monitoring and in-vehicle sensing, where high temporal resolution and minimal identity leakage are beneficial. Finally, we reconstruct detailed 3D facial meshes from single 2D frames using 2D-to-3D morphing techniques to capture fine-grained structural details. This modality enhances geometric understanding and supports applications in AR/VR and affective computing. To validate the robustness and utility of the generated dataset, we perform a comprehensive evaluation using Kernel Inception Distance (KID), BRISQUE, CLIP score, and identity similarity metrics. We further assess downstream applicability employing the state-of-the-art facial expression classification networks and event facial landmarks tests for downstream machine learning tasks. As a key contribution, we open-source a large-scale, multimodality, multi-race adult dataset, enabling future research in secure and ethically grounded synthetic data for facial biometrics and facial analysis applications. The project website, along with the complete adult multimodality dataset and the fine-tuned model, is available at (https://mali-farooq.github.io/SynAdult/).

Ensuring the availability of child facial datasets is essential for advancing AI applications, yet legal, ethical, and data scarcity concerns pose significant challenges. Current generative models such as StyleGAN excel at producing synthetic facial data but struggle with temporal consistency, control over output attributes, and diversity in rendered features. These limitations underscore the need for a more robust and adaptable framework. In this research, we propose the ChildDiffusion framework, designed to generate photorealistic child facial data using diffusion models. The framework integrates intelligent augmentations via short text prompts, employs various image samplers, and leverages ControlNet for enhanced model conditioning. Additionally, we have used large language models (LLMs) to provide complex textual guidance to enable precise image-to-image transformations, facilitating the curation of diverse, high-quality datasets. The model was validated by generating child faces with varied ethnicities, facial expressions, poses, lighting conditions, eye-blinking effects, accessories, hair colors, and multi-subject compositions. To exemplify its potential, we open-sourced a dataset of 2.5k child facial samples across five ethnic classes, which underwent rigorous qualitative and quantitative evaluations. Further, we fine-tuned a Vision Transformer model to classify child ethnicity as a downstream task, demonstrating the framework’s utility. This research advances generative AI by addressing data scarcity and ethical challenges, showcasing how diffusion models can produce realistic child facial data while ensuring compliance with privacy standards. The versatile ChildDiffusion framework offers broad potential for machine learning applications, serving as a valuable tool for AI innovation. The project website, along with the complete ChildRace dataset and the fine-tuned model, is available at (https://mali-farooq.github.io/childdiffusion/).

Establishing the identity of an individual from their facial data is widely adopted across the consumer sector, driven by the use of facial authentication on handheld devices. This widespread use of facial authentication technology has raised other issues, in particular those of biases in the underlying algorithms. Initial studies focused on ethnic or gender biases, but another area is that of age-related biases. This research work focuses on the challenge of face recognition over decades-long time intervals and explores the feasibility of utilizing synthetic ageing data to improve the robustness of face recognition models in recognizing people across these longer time intervals. To achieve this, we first design a set of experiments to evaluate state-of-the-art synthetic ageing methods. In the next stage, we explore the effect of age intervals on a reference face recognition algorithm using both synthetic and real ageing data to perform rigorous validation. We then use these synthetic age data as an augmentation method to facilitate the age-invariant face recognition algorithm. Extensive experimental results demonstrate a notable improvement in the recognition rate of the model trained on synthetic ageing images, with an increase of 3.33% compared to the baseline model when tested on images with a 40-year age gap. Additionally, our models exhibit competitive performance when validated on benchmark cross-age datasets and general face recognition datasets. These findings underscore the potential of synthetic age data to enhance the performance of age-invariant face recognition systems.

Heart rate is a crucial metric in health monitoring. Traditional computer vision solutions estimate cardiac signals by detecting physical manifestations of heartbeats, such as facial discoloration caused by blood oxygenation changes, from subject videos using regression methods. As continuous signals are more complex and expensive to de-noise, this study introduces an alternative approach, employing end-to-end classification models to remotely derive a discrete representation of cardiac signals from face videos. These visual cardiac signal classifiers are trained on discretized cardiac signals, a novel pre-processing method with limited precedent in health monitoring literature. Consequently, various methods to convert continuous cardiac signals into binary form are presented, and their impact on training is evaluated. An implementation of this approach, the temporal shift convolutional attention binary classifier, is presented using the regression-based convolutional attention network architecture. The classifier and a baseline regression model are trained and tested using publicly available and locally collected datasets designed for heart signal detection from face video. The model performance is then assessed based on the heart rate error from the extracted cardiac signals. Results show the proposed method outperforms the baseline on the UBFC-rPPG dataset, reducing cross-dataset root mean square error from 2.33 to 1.63 beats per minute. However, both models struggled to generalize to the PURE dataset, with root mean square errors of 12.40 and 16.29 beats per minute, respectively. Additionally, the proposed approach reduces the computational complexity of model output post-processing, enhancing its suitability for real-time applications and deployment on systems with restricted resources.