Task assignment is a research focus of Mobile Crowd Sensing(MCS). On the arrival of a sensing task, most researches use the way of immediate recruitment to assign the task, ignoring the impact of the task assignment time point. However, the arrival time and start time of a task are often different, hence an earlier assignment decision leads to rough predictions of users reaching the destination of task, which reduces task completion ratio and budget utilization ratio. In order to solve this problem, in this paper we propose a dynamic delayed-decision task assignment method. Firstly, we formalize a new task assignment problem considering the time point of task assignment, in which each selected user can complete as many tasks as possible within the given spatial-temporal constraints, rather than only one task, and propose a method of decision time point selection using a delayed-decision strategy to deal with the tasks of dynamically arriving at the MCS platform. Secondly, we propose two mobility prediction methods to efficiently compute the probabilities of users reaching the destinations before the end time of the sensing tasks, and then propose the related task assignment algorithms, namely with TDMar and TDMeta, which use the proposed semi-Markov prediction and meta-path prediction, respectively. Finally, by using large-scale real data sets, we evaluate the two algorithms and compare them with the baseline methods. The results show that using delayed-decision strategy could evidently improve the task completion ratio and budget utilization ratio, and also decrease the user singleness.

Object detection is an important task in drone vision. Since the number of objects and their scales always vary greatly in the drone-captured video, small object-oriented feature becomes the bottleneck of model performance, and most existing object detectors tend to underperform in drone-vision scenes. To solve these problems, we propose a novel detector named YOLO-Drone. In the proposed detector, the backbone of YOLO is firstly replaced with ConvNeXt, which is the state-of-the-art one to extract more discriminative features. Then, a novel scale-aware attention (SAA) module is designed in detection head to solve the large disparity scale problem. A scale-sensitive loss (SSL) is also introduced to put more emphasis on object scale to enhance the discriminative ability of the proposed detector. Experimental results on the latest VisDrone 2022 test-challenge dataset (detection track) show that our detector can achieve average precision (AP) of 39.43%, which is tied with the previous state-of-the-art, meanwhile, reducing 39.8% of the computational cost.