Daiqin Yang

3-D BLE Indoor Localization Based on Denoising Autoencoder

Bluetooth low energy (BLE)-based indoor localization has attracted increasing interests for its low-cost, low-power consumption, and ubiquitous availability in mobile devices. In this paper, a novel denoising autoencoder-based BLE indoor localization (DABIL) method is proposed to provide high-performance 3-D positioning in large indoor places. A deep learning model, called denoising autoencoder, is adopted to extract robust fingerprint patterns from received signal strength indicator measurements, and a fingerprint database is constructed with reference locations in 3-D space, rather than traditional 2-D plane. Field experiments show that 3-D space fingerprinting can effectively increase positioning accuracy, and DABIL performs the best in terms of both horizontal accuracy and vertical accuracy, comparing with a traditional fingerprinting method and a deep learning-based method. Moreover, it can achieve stable performance with incomplete beacon measurements due to unpredictable BLE beacon lost.

Adaptive background for real-time visual tracking

Visual tracking plays a fundamental role in many applications, such as video surveillance, image compression and three-dimensional reconstruction. From the perspective of accuracy and complexity, correlation filter for target tracking has been proved to be one of the most efficient algorithms. However, it suffers from some difficulties when tracking complex objects with rotations, occlusions and other distractions. To improve its robustness, in this paper, we propose an adaptive background model (ABM) to realize real-time visual tracking with a high adaptivity and accuracy. We use the static background information to help tracking instead of only focusing on the target itself, especially when there are great appearance changes. Moreover, we use peak to side-lobe ratio to update the ABM. As shown in the experiments, our proposed method achieves effective performance in visual tracking with good tradeoff between computational complexity and accuracy.

Object Tracking over a Multiple-Camera Network

In this paper, we describe our system for object tracking over a multiple-camera network task in BigMM Challenge in conjunction with the first IEEE International Conference on Multimedia Big Data (BigMM 2015). We focus on the detection and tracking of pedestrians and vehicles. Based on background modeling, we use HOG and SVM to detect pedestrian and morphological processing to detect vehicle in single camera then use spatio-temporal local context for robust object tracking. The features and trajectory of each object in the multiple-camera network are analyzed for matching and camera geometric projection is also employed to optimize the trajectory. We also include the trajectory visualization in our GIS based experiments.