Internship Opportunity At E3Da Research Unit - Edge Ai For Birds And Bats Audio Classification [...] Trento

Internship Opportunity At E3Da Research Unit - Edge Ai For Birds And Bats Audio Classification [...]
Trento
Trentino Alto Adige, Trento

Internship Opportunity at E3DA Research Unit - Edge AI for Birds and Bats Audio Classification in Forest Environments

FBK is opening a new internship opportunity in the Energy Efficient Embedded Digital Architectures - E3DA research unit of DigiS research center.

The E3DA research unit focuses on wireless resource-constrained embedded technologies with a system level approach targeting energy efficiency and Artificial Intelligence (AI) embedding at the very edge.

The activity spans from hardware-software development for low-power wireless smart sensing devices to power management techniques, low power multi-hop wireless protocols, and on-board processing in resource-limited devices with a focus on Edge AI algorithms on multiple sensor data types (e. g.

Images, Audio, Biosignals and others).

The goal of this project is to design, develop, and test a Bird and Bat Recognition System able to run on acoustic loggers used for environmental monitoring.

It offers hands-on experience with AI, embedded systems, and energy-efficient technologies.

The internship will be customized to suit the candidate's university major, research interests, and strengths.

Project Highlights: Efficient Neural Network Design and Training : Design and Train a model for bird sound recognition inspired by Birdnet tailored for low power microcontrollers.

This includes employing distillation techniques, reducing the number of classes to significant bird species and exploiting Efficient Neural Network Architecture Design Techniques.

Deployment on Embedded Systems : Implement the model on a full-spectrum acoustic logger (AudioMoth ).

Performance Benchmarking : Analyze energy efficiency and inference time on the device.

Signal Processing and Communication : Study the interplay between the recognition system and the acoustic logging system utilizing different Preprocessing Chains and wireless transmission of the resulting analysis.

Possible Activities: Development of an efficient neural network model for bird sound recognition.

Deployment of the developed model on an acoustic logger.

Benchmarking energy efficiency and optimizing performance.

Exploring low-power wireless communication for acoustic loggers.

Required Background: We are looking for motivated candidates with the following qualifications:

Currently enrolled in a Master's program in Computer Engineering, Telecommunications, Electrical Engineering, or Computer Science.

Understanding of Deep Learning Operators and how they are implemented in popular Deep Learning tools and frameworks.

Familiarity with the PyTorch framework.

A solid grasp of signal processing concepts, particularly in audio signal processing.

Additional Requirements: Experience with programming in C/C++.

Experience with embedded systems and/or TinyML.

Additional Opportunities: This internship may be extended into a full research thesis for interested candidates.

We offer: 3-6 months, depending on the candidate's needs and preparation.

Canteen (except for UniTN students).

Support for the search for accommodation at the affiliated structures (no allowance).

#J-18808-Ljbffr