Leonardo W. Estevez

Real-time optimization of Viola -Jones face detection for mobile platforms

Face detection algorithms based on the Viola-Jones object detection approach is widely adopted in digital camera products. Due to the computational complexity of these algorithms, often a hardware coprocessor is used for their real-time operation. This paper discusses how to achieve real-time software-based implementation of these algorithms on mobile devices that have relatively limited processing and memory capabilities. Various optimization techniques are discussed and an example implementation outcome on an actual mobile platform is presented.

Empowering Healthcare Patients with Smart Technology

Continua Health Alliance is developing a set of technical interoperability guidelines for personal telehealth systems that will make it possible to better monitor high-yield microevents over time and thereby enable timely lifestyle adjustments.

Energy Efficient Contextual Sensing for Elderly Care

Technology assisted safe living has great potential in revolutionizing the way healthcare is provided to elderly and needy population. A wireless sensor network (WSN) based system for sensing and reporting events based on context is presented in this paper. It is demonstrated that by proper use of architectures for supporting WSNs and by exploiting very recent technological advances, it is now possible to build and deploy extremely energy efficient systems with very long and dependable battery life. The system evolves over most existing approaches by highly localizing the computations for detecting events and transmitting only positive events where care providers may need to be alerted.

Suitability of NFC for Medical Device Communication and Power Delivery

Near Field Communications (NFC) is a 13.56 MHz inductively coupled power delivery and communication protocol that extends the ISO 14443 RFID standard. Low cost NFC scanner subsystems are anticipated to be widely incorporated in coming generations of commodity cellular phones. We consider the potential of this emerging infrastructure to provide convenient and low cost power distribution and communication channels for a range of medical devices. For example, an NFC device within a cell phone could relay measurements collected from a defibrillator-pacemaker to a monitoring physician, remotely control an insulin pump, or activate an implanted neural simulation system. NFC devices pose similar bio-compatibility challenges to other implanted electronics without requiring the provisioning of battery power to support communication. Furthermore, an NFC communication subsystems power-independence provides a measure of defense against potential denial-of-service attacks that consume power in order to discharge a capacity-limited power source. The 13.56 MHz band has minimal interaction with human and animal tissues. We conducted several successful proof-of- concept experiments communicating with with ISO 14443 tags implanted at multiple locations within a human cadaver. Magnetic field strength decays with the cube of distance-to- antenna, limiting limits the range of potential eavesdroppers. At present, NFC protocols do not provide an appropriate set of privacy properties for implanted medical applications. However, NFC devices are implemented using embedded general purpose processors and thus only software modifications would be required to support protocol extensions with enhanced privacy.

Real-time American Sign Language Recognition using wrist-worn motion and surface EMG sensors

A Sign Language Recognition (SLR) system enables communication between hearing disabled individuals and those who can hear and speak. With the prevalence of the wearable computers, this technology is becoming an important human computer interface capable of reading hand gestures and inferring user;s intent. In this paper, we propose a real-time American SLR system leveraging fusion of surface electromyography (sEMG) and a wrist-worn inertial sensor at the feature level. A feature selection is provided for 40 most commonly used words and for four subjects. The experimental results show that after feature selection and conditioning, our system achieves 95.94% recognition rate. The results also illustrate the fusion of two modalities perform better than using only the inertial sensor. We observed that only one channel of sEMG (out of four) located on the wrist and under the wrist-watch is sufficient.

Power efficient multi-band contextual activity monitoring for assistive environments

The advancement in technology for patient monitoring and smart living environments is making a huge impact on providing health care for the elderly and needy population. An energy efficient context aware system for sensing and reporting events based on wireless sensor networks (WSN) is presented in this paper. It is demonstrated that it is possible to build and deploy power efficient systems with an extended battery life by exploiting the recent technological innovations and properly using the architectural framework for supporting WSNs. The system evolves with highly localized computations aided by the contextual information which helps in detecting events and alerting the care provider for only the positive events.

Energy-efficient ambulatory activity monitoring for disease management

Technology assisted health and disease management has a potential of producing high quality life in various segments of population. A wireless sensor network (WSN) based system for sensing and reporting ambulatory activity for patients is presented in this paper. It is demonstrated that by proper use of architectures for supporting WSNs and by exploiting very recent technological advances, it is now possible to build and deploy extremely energy efficient systems with very long and dependable battery life. The system evolves over most existing approaches by highly localizing the computations for detecting events and intermittently transmitting data related to activities. The paper describes design of the system and analysis for energy efficiency.

Real-time video surveillance system architecture

This paper presents some approaches intended to maximize the available processor bandwidth of a real time video surveillance system. The techniques approach this goal from data centric and process centric perspectives. Data vectorization focuses on organizing and transformign data to more efficiently process it. Some cache considerations and compressed data analysis techniques are therefore reviewed. Dynamic scheduling focuses on using application specific information to reduce the iterations are complexity of repeated processes. Some novel applications of these techniques to video tracking gand recognition are presented. Some implementation examples are also provided indicating that the trade-offs in such an implementation are economically viable.

Real-Time implementation of logo detection on open source BeagleBoard

This paper presents the real-time implementation of our previously developed logo detection and tracking algorithm on the open source BeagleBoard mobile platform. This platform has an OMAP processor that incorporates an ARM Cortex processor. The algorithm combines Scale Invariant Feature Transform (SIFT) with k-means clustering, online color calibration and moment invariants to robustly detect and track logos in video. Various optimization steps that are carried out to allow the real-time execution of the algorithm on BeagleBoard are discussed. The results obtained are compared to the PC real-time implementation results.