To address privacy concerns regarding digital video surveillance cameras, we propose a practical, real-time approach that preserves the ability to observe actions while obscuring individual identities. In the Respectful Cameras system, people who wish to remain anonymous wear colored markers such as hats or vests. The system automatically tracks these markers using statistical learning and classification to infer the location and size of each face. It obscures faces with solid ellipsoidal overlays, while minimizing the overlay area to maximize the remaining observable region of the scene. Our approach uses a visual color-tracker based on a nine dimensional color-space using a Probabilistic Adaptive Boosting (AdaBoost) classifier with axis-aligned hyperplanes as weak hypotheses. We then use Sampling Importance Resampling (SIR) Particle Filtering to incorporate interframe temporal information. Because our system explicitly tracks markers, our system is well-suited for applications with dynamic backgrounds or where the camera can move (e.g. under remote control). We present experiments illustrating the performance of our system in both indoor and outdoor settings, with occlusions, multiple crossing targets, lighting changes, and observation by a moving robotic camera. Results suggest that our implementation can track markers and keep false negative rates below 2%. Fig. 1 A sample video frame is on left. The system has been trained to track green vests such as the one worn by the man with the outstretched arm. The system output is shown in the frame on the right, where an elliptical overlay hides the face of this man. The remainder of the scene including faces of workers not wearing green vests, remain visible. Note how the system successfully covers the face even when the vest is subjected to a shadow and a partial occlusion. Please visit "http://goldberg.berkeley.edu/RespectfulCameras" for more examples including video sequences.
To address privacy concerns regarding digital video surveillance cameras, we propose a practical, real-time approach that preserves the ability to observe actions while obscuring individual identities. In the Respectful Cameras system, people who wish to remain anonymous wear colored markers such as hats or vests. The system automatically tracks these markers using statistical learning and classification to infer the location and size of each face. It obscures faces with solid ellipsoidal overlays, while minimizing the overlay area to maximize the remaining observable region of the scene. Our approach uses a visual color-tracker based on a nine dimensional color-space using a Probabilistic Adaptive Boosting (AdaBoost) classifier with axis-aligned hyperplanes as weak hypotheses. We then use Sampling Importance Resampling (SIR) Particle Filtering to incorporate interframe temporal information. Because our system explicitly tracks markers, our system is well-suited for applications with dynamic backgrounds or where the camera can move (e.g. under remote control). We present experiments illustrating the performance of our system in both indoor and outdoor settings, with occlusions, multiple crossing targets, lighting changes, and observation by a moving robotic camera. Results suggest that our implementation can track markers and keep false negative rates below 2%. Fig. 1 A sample video frame is on left. The system has been trained to track green vests such as the one worn by the man with the outstretched arm. The system output is shown in the frame on the right, where an elliptical overlay hides the face of this man. The remainder of the scene including faces of workers not wearing green vests, remain visible. Note how the system successfully covers the face even when the vest is subjected to a shadow and a partial occlusion. Please visit "http://goldberg.berkeley.edu/RespectfulCameras" for more examples including video sequences.
Abstract-We consider the problem of tracking multiple moving robots using noisy sensing of inter-robot and interbeacon distances. Sensing is local: there are three fixed beacons at known locations, so distance and position estimates propagate across multiple robots. We show that the technique of Nonparametric Belief Propagation (NBP), a graph-based generalization of particle filtering, can address this problem and model multi-modal and ring-shaped uncertainty distributions. NBP provides the basis for distributed algorithms in which messages are exchanged between local neighbors. Generalizing previous approaches to localization in static sensor networks, we improve efficiency and accuracy by using a dynamics model for temporal tracking. We compare the NBP dynamic tracking algorithm with SMCL+R, a sequential Monte Carlo algorithm [1]. Whereas NBP currently requires more computation, it converges in more cases and provides estimates that are 3 to 4 times more accurate. NBP also facilitates probabilistic models of sensor accuracy and network connectivity.
Abstract-Our objective is to automatically track and capture photos of an intruder using a robotic pan-tiltzoom camera. In this paper, we consider the problem of automated position estimation using a wireless network of inexpensive binary motion sensors. The challenge is to incorporate data from a network of noisy sensors that suffer from refractory periods during which they may be unresponsive. We propose an estimation method based on Particle Filtering, a numerical sequential Monte Carlo technique. We model sensors with conditional probability density functions and incorporate a probabilistic model of an intruder's state that utilizes velocity. We present simulation and experiments with passive infrared (PIR) motion sensors that suggest that our estimator is effective and degrades gracefully with increasing sensor refractory periods.
Large-scale sensor network applications require in-network processing and data fusion to compute statistically relevant summaries of the sensed measurements. This paper studies distributed message-passing algorithms, in which neighboring nodes in the network pass local information relevant to a global computation, for performing statistical inference. We focus on the class of reweighted belief propagation (RBP) algorithms, which includes as special cases the standard sum-product and max-product algorithms for general networks with cycles, but in contrast to standard algorithms has attractive theoretical properties (uniqueness of fixed points, convergence, and robustness). Our main contribution is to design and implement a practical and modular architecture for implementing RBP algorithms in real networks. In addition, we show how intelligent scheduling of RBP messages can be used to minimize communication between motes and prolong the lifetime of the network. Our simulation and Mica2 mote deployment indicate that the proposed algorithms achieve accurate results despite realworld problems such as dying motes, dead and asymmetric links, and dropped messages. Overall, the class of RBP provides provides an ideal fit for sensor networks due to their distributed nature, requiring only local knowledge and coordination, and little requirements on other services such as reliable transmission.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.