More Like this

1. LowKey: Leveraging Adversarial Attacks to Protect Social Media Users from Facial Recognition

   Cherepanova, Valeriia; Goldblum, Micah; Foley, Harrison; Duan, Shiyuan; Dickerson, John P; Taylor, Gavin; Goldstein, Tom (April 2021, Proceedings of the International Conference on Learning Representations (ICLR))

   Facial recognition systems are increasingly deployed by private corporations, government agencies, and contractors for consumer services and mass surveillance programs alike. These systems are typically built by scraping social media profiles for user images. Adversarial perturbations have been proposed for bypassing facial recognition systems. However, existing methods fail on full-scale systems and commercial APIs. We develop our own adversarial filter that accounts for the entire image processing pipeline and is demonstrably effective against industrial-grade pipelines that include face detection and large scale databases. Additionally, we release an easy-to-use webtool that significantly degrades the accuracy of Amazon Rekognition and the Microsoft Azure Face Recognition API, reducing the accuracy of each to below 1%. 

   more » « less
2. Privacy-Preserving Face Recognition and Verification With Lensless Camera

   https://doi.org/10.1109/TBIOM.2024.3515144  

   Henry, Chris; Salman_Asif, M; Li, Zhu (July 2025, IEEE Transactions on Biometrics, Behavior, and Identity Science)

   Facial recognition technology is becoming increasingly ubiquitous nowadays. Facial recognition systems rely upon large amounts of facial image data. This raises serious privacy concerns since storing this facial data securely is challenging given the constant risk of data breaches or hacking. This paper proposes a privacy-preserving face recognition and verification system that works without compromising the user’s privacy. It utilizes sensor measurements captured by a lensless camera - FlatCam. These sensor measurements are visually unintelligible, preserving the user’s privacy. Our solution works without the knowledge of the camera sensor’s Point Spread Function and does not require image reconstruction at any stage. In order to perform face recognition without information on face images, we propose a Discrete Cosine Transform (DCT) domain sensor measurement learning scheme that can recognize faces without revealing face images. We compute a frequency domain representation by computing the DCT of the sensor measurement at multiple resolutions and then splitting the result into multiple subbands. The network trained using this DCT representation results in huge accuracy gains compared to the accuracy obtained after directly training with sensor measurement. In addition, we further enhance the security of the system by introducing pseudo-random noise at random DCT coefficient locations as a secret key in the proposed DCT representation. It is virtually impossible to recover the face images from the DCT representation without the knowledge of the camera parameters and the noise locations. We evaluated the proposed system on a real lensless camera dataset - the FlatCam Face dataset. Experimental results demonstrate the system is highly secure and can achieve a recognition accuracy of 93.97% while maintaining strong user privacy. 

   more » « less
3. Development of a Facial Feature Based Image Steganography Technology

   https://doi.org/10.1109/CSCI49370.2019.00126  

   Marella, Pranay; Straub, Jeremy; Bernard, Benjamin (December 2019, Proceedings of the 2019 International Conference on Computational Science and Computational Intelligence (CSCI))

   A new image steganography method is proposed for data hiding. This method uses least significant bit (LSB) insertion to hide a message in one of the facial features of a given image. The proposed technique chooses an image of a face from a dataset of 8-bit color images of head poses and performs facial recognition on the image to extract the Cartesian coordinates of the eyes, mouth, and nose. A facial feature is chosen at random and each bit of the binary representation of the message is hidden at the least significant bit in the pixels of the chosen facial feature. © 2019 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. 

   more » « less
4. Triplet loss-based deep learning frameworks for off-angle iris recognition

   https://doi.org/10.1117/12.3055314  

   Tebor, Daniel; Karakaya, Mahmut (May 2025, SPIE)

   Alam, Mohammad S; Asari, Vijayan K (Ed.)

   Iris recognition is a reliable biometric identification method known for its low false-acceptance rates. However, capturing ideal iris images is often challenging and time-consuming, which can degrade the performance of iris recognition systems when using non-ideal images. Enhancing iris recognition performance for non-ideal images would expedite and make the process more flexible. Off-angle iris images are a common type of non-ideal iris images, and converting them to their frontal version is not as simple as making geometric transformations on the off-angle iris images. Due to challenging factors such as corneal refraction and limbus occlusion, frontal projection requires a more comprehensive approach. Pix2Pix generative adversarial networks (GANs), with their pairwise image conversion capability, provide the ideal foil for such a tailored approach. We demonstrate how Pix2Pix GANs can effectively be used for the problem of converting off-angle iris images to frontal iris images. We provide a comprehensive exploration of techniques using Pix2Pix GAN to enhance off-angle to frontal iris image transformation by introducing variations in the loss functions of Pix2Pix GAN for better capturing the iris textures and the low contrast, changing the medium of input from normalized iris to iris codes, and ultimately delving deeper into studying which regions of the Gabor filters contribute the most to iris recognition performance. 

   more » « less
5. Delighting Palates with AI: Reinforcement Learning’s Triumph in Crafting Personalized Meal Plans with High User Acceptance

   https://doi.org/10.3390/nu16030346  

   Amiri, Maryam; Sarani_Rad, Fatemeh; Li, Juan (February 2024, Nutrients)

   Eating, central to human existence, is influenced by a myriad of factors, including nutrition, health, personal taste, cultural background, and flavor preferences. The challenge of devising personalized meal plans that effectively encompass these dimensions is formidable. A crucial shortfall in many existing meal-planning systems is poor user adherence, often stemming from a disconnect between the plan and the user’s lifestyle, preferences, or unseen eating patterns. Our study introduces a pioneering algorithm, CFRL, which melds reinforcement learning (RL) with collaborative filtering (CF) in a unique synergy. This algorithm not only addresses nutritional and health considerations but also dynamically adapts to and uncovers latent user eating habits, thereby significantly enhancing user acceptance and adherence. CFRL utilizes Markov decision processes (MDPs) for interactive meal recommendations and incorporates a CF-based MDP framework to align with broader user preferences, translated into a shared latent vector space. Central to CFRL is its innovative reward-shaping mechanism, rooted in multi-criteria decision-making that includes user ratings, preferences, and nutritional data. This results in versatile, user-specific meal plans. Our comparative analysis with four baseline methods showcases CFRL’s superior performance in key metrics like user satisfaction and nutritional adequacy. This research underscores the effectiveness of combining RL and CF in personalized meal planning, marking a substantial advancement over traditional approaches. 

   more » « less