Skip to content
Imtiaz Hossain

Imtiaz Hossain / Dhaka, Bangladesh/ open to research & roles

Computer VisionNLPDeep LearningResearchProduction Systems
14case studies
88.22%best test dice
27nlp experiments
$0/moinfra cost

01 / selected work

Systems that shipped

03 / research

Obscrowd: Multi-Portrait Biometric Unlinkability under Generative Editing

Generative image editors can modify a portrait (changing pose, expression, background, or style) while leaving enough biometric detail in the edited output that face-recognition systems can still match it back to the original subject, creating a privacy risk that is distinct from image forgery. Existing protective methods typically focus…

preprintWorking paper / 2026
Read the publication →
Original, protected, and the imperceptible perturbation (delta map at plus/minus 0.16). After an instruct-pix2pix smile edit, face-recognition similarity falls from 0.957 (unprotected) to 0.138 (protected).
Original, protected, and the imperceptible perturbation (delta map at plus/minus 0.16). After an instruct-pix2pix smile edit, face-recognition similarity falls from 0.957 (unprotected) to 0.138 (protected).

04 / next experiment

Have a hard problem that needs an intelligent system?