The Face Recognition system has a challenge when the conditions of the input images have differences in quality from the images that have been enrolled in the database. One of the causes is the variation in lighting that causes illumination in image. We used images with normal lighting, as well as four images that have variations in the lighting/illumination directions. Face Image Quality Assessment (FIQA) helps the face recognition system to ensure the optimum captured image quality for enrollment and verification process. We use both supervised (FaceQnet) and unsupervised (SDD-FIQA, SER-FIQ) FIQA method against the Asian Face Image dataset. The result shows that filtering images using FIQA method can reduce FNMR by 58.89% in matching images whose light direction is from below. Images with type 2 illumination, where an image whose light comes from below matched with normal image, gave the lowest result in FRR compared to other types of illumination when tested with 3 FIQA methods.
The development of computer vision involving artificial intelligence has been applied in several sectors. People counting is not only a tool for measuring the number of people but also a key element in security management, compliance, and operational efficiency, while creating a safe and enjoyable environment in public areas. One of them is about single-stage layer-based people counting using You Look Only Once (Yolo) and Single Shoot Detector (SSD). To implement object detection, a backbone for feature extraction is required. In this article, Yolo uses v7 with an Extended Efficient Layer Aggregation Network (ELAN) backbone, while for SSD it uses a mobilenet, inception, and resnet backbone. Data for testing was obtained by taking pictures in one of the public areas, namely at a train station in Indonesia. Evaluation uses average precision (AP) and calculates the accuracy of the number of people. The number of people is calculated based on three categories, small conditions (1-15 people), medium conditions (16-25 people), and many more than 26 people (crowded condition). Based on these results, the most significant value was obtained at 0.985 for YoloV7. The highest accuracy was obtained with a result of 100% for YoloV7 in a small number of conditions (1-15 people).