Food object recognition systems present an attractive and useful research field since they enable objective measurements of eating activity. This feature is helpful and welcome in many dieting related instances, especially for managing health conditions or for analyzing eating patterns of research subjects. We evaluate current food object recognition systems that were implemented on a mobile device. The evaluation was provided by analysing each particular system through its food recognition process. The whole recognition process was divided into 6 distinct stages: image acquisition, image processing, image segmentation, feature extraction, image classification, and volume estimation. Through the analysis, the authors provide a categorization of mobile food recognition systems: recorder systems, suggester systems, and clinical responders. Each group is aimed at a different scenario which helps to identify features a particular system should focus its development on.
F.30 Professional assessment of the situation
COBISS.SI-ID: 1538556099In the article we present a novel, complex and publicly available database for sclera recognition and analysis of the state-of-the-art models. The importance of the group behind this work is evident also from the competition paper about sclera segmentation from 2020 [COBISS.SI-ID 53460483], which we organised at a prestige biometrics conference.
F.15 Development of a new information system/databases
COBISS.SI-ID: 1538534595In the era of COVID-19 the question of correctly worn face masks became very important. The article suggests a deployable solution, which answers to the given question in real-time and with very high reliability. The solution is publicly available. It was also presented as an invited tutorial at an international conference in South Korea.
F.17 Transfer of existing technologies, know-how, methods and procedures into practice
COBISS.SI-ID: 53569795This paper presented a novel two-stage segmentation-based deep-learning architecture designed for detection and segmentation of surface anomalies using fully-supervised learning. The method is trained on input images and the corresponding segmentation masks and outputs a decision for each image whether it contains an error or not, as well as a segmentation mask that localises the visual defect in the image. The method is based on image segmentation in the first stage, which is then upgraded with a classification head to make the final decision. Comparison to other related deep-learning methods, including the state-of-the-art commercial software, showed that the proposed approach outperforms the related approaches. This is demonstrated on a newly created dataset KSDD, based on a real-world quality control case, where the proposed approach achieved excellent results. The research community has shown a large interest in this work, both, in the source code as well as in the image dataset that we have both made publicly available; the latter has already been downloaded more than 6600 times. We also upgraded the proposed method into an end-to-end architecture, and achieved even better results; we were able to completely solve the DAGM and KSDD benchmark datasets, without any false detection. The description of the upgraded method and results were presented at the International Conference on Pattern Recognition 2020. In cooperation with our industry partners, we have also tested the method on real-world practical problems, where it has achieved promising results.
F.13 Development of new production methods and tools or processes
COBISS.SI-ID: 1538225859This paper addressed detection and recognition of a large number of object categories, applied to the problem of traffic-sign sign detection and recognition. A convolutional neural network approach Mask R-CNN, was adapted to address the full pipeline of detection and recognition with automatic end-to-end learning. Several proposed improvements were evaluated on the detection of 200 traffic-sign categories from a newly created dataset. Evaluation focused on highly challenging traffic-sign categories that have not yet been considered in previous works. A comprehensive analysis of the deep learning method for the detection of traffic signs with large intra-category appearance variation showed below 3% error rates for the proposed approach, thus outperforming the related approaches.
F.17 Transfer of existing technologies, know-how, methods and procedures into practice
COBISS.SI-ID: 1538227907