Deep learning-based object detection has seen a surge in applications for sensing systems on mobile devices. In this context, objects are identified and tracked across video frames, facilitating the calculation of associated events of interest. A significant research challenge refers to the acceleration of processing speed, which is constrained by deep learning-based object detection due to its intensive resource requirements. This paper focuses on a typical mobile sensing scenario, wherein sequences of frames containing objects of interest are sparsely dispersed throughout the video stream. Given that many of the frames lack objects, allocating substantial computational resources to detect them becomes inefficient. In light of this, we propose a stochastic scheduling algorithm, JumpQ. JumpQ performs per-frame detection when anticipating the presence of objects in the current frames. Consecutive negative detections prompt a transition to intermittent detection with a probability that undergoes further decay if the negative detection persists until reaching a predefined limit. Upon a positive detection, JumpQ swiftly reverts to per-frame detection and retraces a specific number of previously buffered frames to ensure the inclusion of potentially missed true frames. A comprehensive experimental study using the garbage bag counting technique was conducted to show the efficiency of JumpQ in accelerating the processing speed by nearly 1.92 times while maintaining a negligible impact on sensing accuracy.

One major research problem regarding mobile sensing systems is how to accelerate the processing speed which is bottle-necked by the deep-learning-based object detection. The focus of this paper is directed towards a typical mobile sensing scenario wherein sequences of frames containing interested objects are sparsely dispersed throughout the video stream. In light of this, we propose a stochastic scheduling algorithm named JumpQ. In the case of consecutive negative detections, JumpQ reduces the probability of detection, while in the case of positive detections, JumpQ promptly returns to frame-by-frame detection and retraces the buffered frames to detect the objects. Our experiment reports that the JumpQ algorithm accelerates processing speed by over 100%, all while incurring a negligible impact on sensing accuracy.

The information of daily garbage diposal can be used to develop many appealing applications in smart cities. This poster introduces DeepCounter, an automotive sensing system to providing a finegrained spatio-temporal distribution on the amount of disposed garbage bags. In the system, deep learning based image processing is used to automatically count the number of collected garbage bags from the video taken by a camera mounted on the rear of a garbage truck. A prototype system is implemented and experimental evaluation validates the feasibility of our proposal using realistic garbage collection videos in Fujisawa city Japan.

Can we inspect the road condition at a low cost? City infrastructures, such as roads are very important for citizens to their city lives. Roads require constant inspection and repair due to deterioration, but it is expensive to do so with manual labor. Meanwhile, there are official city vehicles, especially garbage trucks that run through the entire area of a city every day and have cameras to record their driving. When we use these cameras, we can watch roads conditions anytime, anywhere. In our study, we focus on these cameras and attempt detecting the road damage, such as road marking blur. To achieve our goal, we explore the new system in this paper. This system adopts the object detection approach that is end-to-end learning and based on deep neural networks, which propose the blur region candidate and detect whether the road markings are blurred or not all at once. In our experiment, first, we obtain the drive recorder video from sanitation engineer and then annotate them. After annotation, we trained our models and calculate the mean average precision to evaluate our models. As a result, our model performs on our collected dataset.