This paper proposes a novel task-consistency learning method that enables us to train a vacant space detection network (target task) based on the logic consistency with the semantic outcomes from a flow-based motion behavior classifier (source task) in a parking lot. Note that the source task can introduce false detection during task-consistency learning, which implies noisy rewards or supervision. The target network can be trained in a reinforcement learning setting by appropriately designing the reward mechanism upon semantic consistency. We also introduce a novel symmetric constraint to detect corrupted samples and reduce the effect of noisy rewards. Unlike conventional corrupted learning methods that use only training losses to identify corrupted samples, our symmetric constraint also explores the relationship among training samples to improve performance. Compared with conventional supervised detection methods, the main contribution of our work is the ability to learn a vacant space detector via semantic consistency rather than supervised labels. The dynamic learning property allows the proposed detector to be easily deployed and updated in various lots without heavy human loads. Experiments demonstrate that our noisy task consistency mechanism can be successfully applied to train a vacant space detector from scratch.

In this paper, we proposed a novel task-consistency learning method that allows training a vacant space detection network (target task) based on the logic consistency with the semantic outcomes from a flow-based motion behavior classifier (source task) in a parking lot. By well designing the reward mechanism upon semantic consistency, we show the possibility to train the target network in a reinforcement learning setting. Compared with conventional supervised detection methods, this work's main contribution is to learn a vacant space detector via semantic consistency rather than supervised labels. The dynamic learning property may make the proposed detector been deployed and updated in different lots easily without heavy human loads. The experiments show that based on the task consistency rewards from the motion behavior classifier, the vacant space detector can be trained successfully.

In this paper, we introduce an online-learning method to model the property of an office building. Unlike conventional control methods where the building property is modeled via a simulator or through offline learning, our building model is adaptively updated according to the dynamic response of a real environment. Upon the building model for environment prediction, the proposed action agent can control the heating, ventilation, and air conditioning (HVAC) system in a smarter way by scheduling the temperature reference point. To online learn the model and improve the agent, two practical and seldom discussed issues are addressed. The first challenge is data bias where the collected initial training dataset can only partially reveal the statistical mapping between the control input and the environment response. Hence, the trained model may lack generalization. To overcome the data bias issue, a data augmentation method is proposed to embed physical logic in order to train a proper initial model. Next, an online learning process is introduced to update the model generality during the system operation phase. The second practical issue is the constraints on agent exploration for discovering unknown data samples. During the business hours, to comfort employees, a control agent is not allowed to explore the possible controlling space randomly. To balance data collection and control stability, we introduce a hybrid control strategy that considers both the human control rule and the agent action. A confidence score of the agent model is also automatically estimated to determine a suitable control strategy finally. Our experiments have realized in an office building. The results outperform conventional methods and show its superior in terms of control stability.

For human skeleton tracking, conventional methods may directly rely on the Kinect built-in function to extract skeleton joints and realize the tracking function. However, the left-right confusion problem and the self-occlusion problem would make the Kinect joints unstable and incorrect. In this paper, we aim to solve these crucial issues and realize a reliable 3D skeleton tracking system via multiple Kinect cameras. First, the proposed method corrected the Kinect skeleton by referring to the OpenPose-extracted joints. Since OpenPose extracts the joints by analyzing image content, we can differentiate the front side and the back side and thus can correct the left-right confusion. Second, to alleviate the self-occlusion problem, we back-project the 2D OpenPose joints from multiple cameras and form an alternative 3D skeleton. The OpenPose reconstructed joints are treated as robust 3D anchors for multiple skeleton fusion. Last but not least, we introduce the inter-joint constraints into our skeleton tracking framework so that we can trace all joints simultaneously, make sure the skeleton movement consistent, and well maintain the length between neighboring joints. We evaluate our method with challenging actions. A practical system is also built for demonstration. The experimental results show the system can track skeleton stably without error propagation and vibration. The average localization error is also smaller than conventional methods.

Inspecting X-ray images is an essential aspect of medical diagnosis. However, due to an X-ray's low contrast and low dynamic range, important aspects such as organs, bones, and nodules become difficult to identify. Hence, contrast adjustment is critical, especially because of its ability to enhance the details in both bright and dark regions. For X-ray image enhancement, we therefore propose a new concept based on component attenuation. Notably, we assumed an X-ray image could be decomposed into tissue components and important details. Since tissues may not be the major primary focus of an X-ray, we proposed enhancing the visual contrast by adaptive tissue attenuation and dynamic range stretching. Via component decomposition and tissue attenuation, a parametric adjustment model was deduced to generate many enhanced images at once. Finally, an ensemble framework was proposed for fusing these enhanced images and producing a high-contrast output in both bright and dark regions. We have used measurement metrics to evaluate our system and achieved promising scores in each. An online testing system was also built for subjective evaluation. Moreover, we applied our system to an X-ray data set provided by the Japanese Society of Radiological Technology to help with nodule detection. The experimental results of which demonstrated the effectiveness of our method.

Abstract During the past decades, many fingerprint‐based indoor positioning systems have been proposed and have achieved great success. However, uncontrolled effects of device diversity, signal noise, and dynamic obstacles could recognizably degrade the performance of modern fingerprint‐based indoor localization systems. In this paper, to amend the variations in radio signal strengths (RSSs) caused by device diversity, we proposed an automatic device calibration process. Because of device diversity, the sensed RSS would deviate from the trained radio map and thus leads to poor positioning. An RSS transform function could be adopted to calibrate the RSS variation between different devices and overcome the device diversity problem. However, to train the transform function, a data collection process is required. Unlike conventional calibration methods requiring manual data collection, we proposed a landmark‐based automatic collection process. Based on the detection of Wi‐Fi landmarks, our system could automatically collect pair‐wise RSS samples between devices and train the RSS transform function without extra human power. In addition, to well represent the effects of signal noise and dynamic obstacles, a region‐based RSS modeling method was also proposed. The proposed modeling method allows our system to perform region‐based target localization and utilize more robust region information for localization. Experiments in various environments demonstrate that our system could give a better positioning performance by properly handling the RSS variation caused by signal noise, dynamic environment, and device diversity. Copyright © 2015 John Wiley & Sons, Ltd.

X-ray imaging is an efficient tool for health inspection. The energy of received X-ray could reveal the density inside human body and be represented in an X-ray image. In general, the bright regions of an X-ray image are of interest since most important matters compactly locating in those regions. However, the low contrast property of the bright regions makes inspectors hard to tell the details. Hence, contrast enhancement of X-ray images especially in the bright regions becomes critical. In the paper, unlike previous algorithms which work well in the dark regions, we thus proposed a tissue attenuation method to enhance the bright regions. We assume the X-ray intensity is composed of tissues and other important details. By locally attenuating the ratios of tissues over the image, we could adaptively enhance the important details especially in the bright regions. To adjust the ratios, a two-step procedure was proposed. First, the tissue component was separated from a given image based on local contrast maximization. Second, an attenuation adjustment was performed to control the ratio of removable tissues in order to correctly enhance contrast. Experimental results also demonstrate the effectiveness of our method.