Keywords: Illicit Drug Abuse Faces; Vulnerability of Face Recognition; Durg Abuse Face Detection
TL;DR: Banchmark the effectiveness and robustness of CNNs for drug abuse face detection
Abstract: Face recognition is one of the secure mediums to access various security-restricted areas such as border control and mobile unlocking. However, face recognition can be severely impacted due to several factors including illicit drug abuse on the facial regions. These abuses drastically alter the appearance of the faces and hence lead to the poor performance of the face recognition algorithms. However, due to the limited availability of the datasets, the research in this field is still in the novice stage. To advance the research, in this research, `we have collected drug abuse face images and proposed a benchmark study to identify whether the face in question is clean or drug abused'. \textit{Further, we have performed the robustness study of detection networks by altering the images by adopting several enhancement filters popular to use before uploading the face images on social media platforms}.
Submission Number: 97
Loading