Project ID
Topic/description
Team
Project: 28
Breaking down a Deep learning model for intermittent operations on an energy-harvesting embedded system In this project you develop a tool to break down any given model into smaller chunks with breakpoints. These breakpoints should enable a microcontroller to do inference with intermittent wake up and sleep cycles.
--
Project: 27
Federated learning for distribute training of ML model in a decentralized way taking care of data privacy Mobile devices often hold only limited data, which is not sufficient to locally train DNN. Instead of offloading the data to cloud for training, Federated learning enables you to train the model in a decentralized way. In this project you develop this pipeline for an application of your choice.
--
Project: 26
Drink spike detection using visible light sensing This project develops a cup holder that can attach to any transparent glass and can monitor if the drink has been spiked. The system is built on a microcontroller using LEDs and photodiodes.
--
Project: 25
Speech liveness detection by fusing multiple sensors People can fake voice of a person on a phone-call or while using personal digital assistants. Can we develop a way to know if the speech is from a real person?
--
Project: 24
Daily activity classification using subsampled acoustic signals This projects develops a real-time daily activity detection system using the captured sound. The sound is subsampled to reduce the input size of the network and fit the model in a microcontroller.
--
Project: 23
Vibro-acoustic Speech Recognition The smartphone may sense both sound and vibration during a phone call. This project will leverage both these channels for better speech recognition in a noisy environment.
--
Project: 22
Localizing and Tracking Passive RFID Tags
This project will use multiple reader antennas to localize the 2D position of an RFID tag and will track a moving tag over time.--
Project: 21
Head Gesture Analysis using smart earphones and its Applications
This project will use sensors in smart earphones to determine head gestures and leverage this information for applications in eating behavior analysis, active listening, long-term sedentary work, and so on.--
Project: 20
Direction of Arrival (DoA) Estimation Using Tiny Array
This project will implement an improvised MUSIC algorithm to estimate high-accuracy DoA of sound signals using a small microphone array. The total aperture of the array is less than a thousandth of the target wavelength.--
Project: 19
Traffic Analytics through Sound Analysis
This project will use an acoustic array for spatial sound analysis at a traffic signal. This will provide various information about the traffic situation at that place.--
Project: 18
Acoustic Scene Analysis from the Public Videos
On social media, users post videos of the same event taken from different locations. This project will leverage sound taken from these videos for acoustic scene analysis.--
Project: 17
Soft Tissue Imaging using Visible Light
This project will apply tomography techniques with visible light to find any tumor under soft tissues in the human body.--
Project: 16
Hand-held ultrasound device for bone-fracture detection
This project will use 2 MHz ultrasound to detect hairline bone fractures.--
Project: 15
Preventing Inaudible Attacks on Voice Assistants
This project will use active techniques to prevent ultrasound attacks (BackDoor) on the voice assistants (Amazon Echo, Google Home).--
Project: 14
Interaction between Smart-car and Pedestrians using DSRC/5G protocol
This project will develop a secure and agile method to communicate between smart-cars and pedestrians to ensure safe co-existence.--
Project: 13
Detect Breathing Pattern Using Wearable Sensors
This project will use wearable acoustic/vibration sensors to detect the users' breathing pattern and will run analytics on this.--
Project: 12
Smart Exoskeleton for Hands
This project will develop an assistive exoskeleton for hands using posture detection and micro-prediction techniques.--
Project: 11
Infants' activity/sleep/breathing monitoring using non-invasive sensors This project will use microphone and motion sensors on the crib to infer various activities of the baby.
--
Project: 10
High-speed screen-to-camera data communication This project will use modulated pixels on the computer's screen to establish high data-rate communication through the smartphone's camera.
--
Project: 09
Privacy-preserving sensing for edge computing This project will develop methods that enable continuous monitoring of daily activities at network edges while ensuring user's privacy.
--
Project: 08
Acoustic tomography for detecting 3D shape on an object This project will use multiple acoustic transmitters and receivers around an object and apply tomography techniques to find the 3D shape of the object.
--
Project: 07
Fine-grained hand gesture recognition using ultrasound reflections This project will use a bed of ultrasound transducers for fine-grained hand gesture recognition. The transducer-bed will transmit signals and observe the reflections for gesture recognition.
--
Project: 06
Charting the interior shape of a cave/tunnel using drone mounted-sensor This project will develop low-power sensing techniques to create a model of a closed region using sensors mounted on a drone.
--
Project: 05
Ultrawideband acoustic sensing This project will develop methods to use the acoustic frequency band from DC to 100kHz.
--
Project: 04
Tracking UWB tags This project will develop computationally efficient methods for tracking tags using ultra-wide-band radio.
--
Project: 03
Acoustic virtual reality This project will use smart headphones to track head motion and will play the binaural sound accordingly to create an illusion of the 3D acoustic scene.
--
Project: 02
Passive structure based low-power gesture-recognition This project will use a 3d-printed structure and ultrasound transducer for hand gesture recognition. The 3d structure will embed direction specific signature to the reflected signals which will help in gesture recognition.
--
Project: 01
Active noise cancellation in earphones This project will use the earphones/headphones as an array or sound sensors. The array will beamform towards the user and use the harmonic structure of speech signal to improve the noise cancellation.
--