Shaun Ahmadian

Heartbeat of a nest: Using imagers as biological sensors

We present a scalable end-to-end system for vision-based monitoring of natural environments, and illustrate its use for the analysis of avian nesting cycles. Our system enables automated analysis of thousands of images, where manual processing would be infeasible. We automate the analysis of raw imaging data using statistics that are tailored to the task of interest. These “features” are a representation to be fed to classifiers that exploit spatial and temporal consistencies. Our testbed can detect the presence or absence of a bird with an accuracy of 82%, count eggs with an accuracy of 84%, and detect the inception of the nesting stage within a day. Our results demonstrate the challenges and potential benefits of using imagers as biological sensors. An exploration of system performance under varying image resolution and frame rate suggest that an in situ adaptive vision system is technically feasible.

HERMES: Mobile system for instability analysis and balance assessment

We introduce Hermes, a lightweight smart shoe and its supporting infrastructure aimed at extending gait and instability analysis and human instability/balance monitoring outside of a laboratory environment. We aimed to create a scientific tool capable of high-level measures, by combining embedded sensing, signal processing and modeling techniques. Hermes monitors walking behavior and uses an instability assessment model to generate quantitative value with episodes of activity identified by physician, researchers or investigators as important. The underlying instability assessment model incorporates variability and correlation of features extracted during ambulation that have been identified by geriatric motion study experts as precursor to instability, balance abnormality and possible fall risk. Hermes provides a mobile, affordable and long-term instability analysis and detection system that is customizable to individual users, and is context-aware, with the capability of being guided by experts. Our experiments demonstrate the feasibility of our model and the complimentary role our system can play by providing long-term monitoring of patients outside a hospital or clinical setting at a reduced cost, with greater user convenience, compliance and inference capabilities that meet the physicians or investigators needs.

Exploring Tradeoffs in Accuracy, Energy and Latency of Scale Invariant Feature Transform in Wireless Camera Networks

Advances in DSP technology create important avenues of research for embedded vision. One such avenue is the investigation of tradeoffs amongst system parameters which affect the energy, accuracy, and latency of the overall system. This paper reports work on benchmarking the performance and cost of scale invariant feature transform (SIFT) for visual classification on a Blackfin DSP processor. Through measurements and modeling of the camera sensor node, we investigate system performance (classification accuracy, latency, energy consumption) in light of image resolution, arithmetic precision, location of processing (local vs. server-side), and processor speed. A case study on counting eggs during avian nesting season is used to experimentally determine the tradeoffs of different design parameters and discuss implications to other application domains.

Network of cyclops; image inference and interpretation in sensor network

Recent technological progress in integrated low power CMOS based imaging devices has led to new type of sensors such as Cyclops. Cyclops is a CMOS image sensor, with reduced complexity and power that allows it to mate with typical sensor network nodes such as Motes. This motivates a new class of sensor networks which exploit vision.In this demonstration we introduce a network of Motes that carry Cyclops sensors. In this network, each individual node can be programmed to perform specific operation on the image. These operations include detecting objects in the scene, detecting edges, reporting histogram of the image, getting the image across the network or getting only particular region of interest.The demonstration showcases the functionality of our network to detect objects. Each node in the network is programmed to detect presence of the objects in its field of view. In addition, users can request the whole image or particular region of interest which the presence of the object has been detected. Performance studies as well as architectural choices will be presented along with outstanding challenges and opportunities.