As the Internet-of-Things (IoT) and edge computing have been major paradigms for distributed data collection, communication, and processing, smart city applications in the real world tend to adopt IoT and edge computing broadly. Today, more and more machine learning algorithms would be deployed into front-end sensors, devices, and edge data centres rather than centralised cloud data centres. However, front-end sensors and devices are usually not so capable as those computing units in huge data centres, and for this sake, in practice, engineers choose to compromise for limited capacity of embedded computing and limited memory, e.g., neural network models being pruned to fit embedded devices. Visual object tracking is one of many important elements of a smart city, and in the IoT and edge computing context, high requirements to computing power and memory space severely prevent massive and accurate tracking. In this paper, we report on our contribution to object tracking on lightweight computing including (1) using limited computing capacity and memory space to realise tracking; (2) proposing a new algorithm region proposal correlation filter fitting for most edge devices. Systematic evaluations show that (1) our techniques can fit most IoT devices; (2) our techniques can keep relatively high accuracy; and (3) the generated model size is much less than others.
In this paper, a simplified constellation architecture consisting of only two navigation satellites located around the Earth-Moon L1,2 libration points is obtained for the Universe Lighthouse. In order to determine the feasible constellations that can achieve continuous global coverage for lunar orbits, an exhaustive search over all possible combinations of libration point orbits is performed first. With the use of a fitting procedure, amplitude relations between the feasible L1 and L2 libration point orbits are derived by polynomial interpolation. After that, a cislunar navigation simulation is conducted to verify the navigation performance of the candidate two-satellite constellations. The final Monte Carlo simulation results indicate that the simplified system is available for cislunar navigation and the best accuracy of a few tens of metres can be achieved for both the trans-lunar cruise phase and lunar orbit phase.
As one of the key components in many computer vision applications, visual tracking has been widely investigated in recent years. Correlation filter-based trackers have achieved compelling performance in terms of both efficiency and accuracy by speeding up online training and inference procedures in the frequency domain. For a visual object, such trackers often perform tracking by searching the most correlated patch within a limited search region in a new frame. The acknowledged problem is that a small search region prohibits the successful tracking of fast moving targets while a large search region increases the difficulty of model learning. In this paper, we propose to use different candidate search regions to train multiple tracking experts to alleviate this dilemma. Based on these experts, the final tracker learns to make the optimal decision using a decision-theoretic online learning approach. Besides, we also establish multi-scale experts to handle scale variations during tracking. The experimental results on the public datasets demonstrate that the proposed approach achieves favorable performance against several state-of-the-art methods.