The performance of learning-based object detection algorithms, which attempt to both classify and locate objects within images, is determined largely by the quality of the annotated dataset used for training. Two types of labelling noises are prevalent: objects that are incorrectly classified (categorization noise) and inaccurate bounding boxes (localization noise); both noises typically occur together in large-scale datasets. In this paper we propose a distillation-based method to train object detectors that takes into account both categorization and localization noise. The key insight underpinning our method is that the early-learning phenomenon - in which models trained on noisy data with mixed clean and false labels tend to first fit to the clean data, and memorize the false labels later -- manifests earlier for localization noise than for categorization noise. We propose a method that uses models from the early-learning phase (before overfitting to noisy data occurs) as a teacher network. A plug-in module implementation compatible with general object detection architectures is developed, and its performance is validated against the state-of-the-art using PASCAL VOC, MS COCO and VinDr-CXR medical detection datasets.

By the wide range of researches on Wireless Sensor Networks (WSNs), various routing schemes with diverse criterion have been introduced. However, techniques that consider the characteristics of the sensed data as definitive parameters in determining the behavior of WSNs are extremely rare. In most monitoring applications, sampled data are not equal in their importance. Unexpected data is more likely to be important. That is, data collected during abrupt changes in the environment are more critical than the others. In this paper, we propose Priority-Based Hybrid Routing (PHR) that provides functions ranging from data priority verification to differentiated services according to different priorities. Data priorities, or importance, are determined by their distinctiveness in relation to past data. Dixon's Test, a hypothesis testing method, is adopted for the process. To provide more reliable routing method for high priority data, PHR offers them a novel, diffusion-based forwarding scheme, referred to as Geographic Diffusion. On the other hand, low priority data are delivered by a famous single path routing algorithm, Ad hoc On-demand Distance Vector (AODV). Furthermore, AODV is revised to compensate with the increased traffic caused by Geographic Diffusion. The performance of PHR is evaluated through various simulations.

In ubiquitous network, node location information is the most essential data for efficiently managing increased mobility and supporting location based services. Therefore, we propose a next generation location information based network, which uses address containing location information. Finally, its functions, effectiveness, and feasibility will be verified through this demonstration.