Machine Learning on Microcontrollers

dc.contributor.advisorZilizi, Gyula
dc.contributor.authorGerelmaa, Tuguldur
dc.contributor.departmentDE--Természettudományi és Technológiai Kar--Fizikai Intézet
dc.date.accessioned2024-12-20T08:16:57Z
dc.date.available2024-12-20T08:16:57Z
dc.date.created2024-11-22
dc.description.abstractMachine Learning has made remarkable progress in the past decade, but current models are computationally expensive. This thesis focuses on deploying machine learning models on microcontrollers. MobileNet based model was trained for two projects: A person detection model was trained, achieving 78.6% accuracy with 225 KB size and 170 KB RAM usage on an ESP32-CAM board. The board sends detected images to a webserver for further analysis. In an another project, keyword detection model was trained, achieving 91.12% accuracy with 120 KB size and 38 KB RAM usage, deployed on a custom made ESP32-based PCB with a MEMS microphone.
dc.description.courseElectrical Engineering
dc.description.degreeBSc/BA
dc.format.extent53
dc.identifier.urihttps://hdl.handle.net/2437/384080
dc.language.isoen
dc.rights.accessHozzáférhető a 2022 decemberi felsőoktatási törvénymódosítás értelmében.
dc.subjectmachine learning
dc.subjectmicrocontroller
dc.subjectwake word
dc.subjectconvolutional neural network
dc.subject.dspaceEngineering Sciences::Electrical Engineering
dc.titleMachine Learning on Microcontrollers
Fájlok
Eredeti köteg (ORIGINAL bundle)
Megjelenítve 1 - 1 (Összesen 1)
Nincs kép
Név:
thesis-fin.pdf
Méret:
6.82 MB
Formátum:
Adobe Portable Document Format
Leírás:
Engedélyek köteg
Megjelenítve 1 - 1 (Összesen 1)
Nincs kép
Név:
license.txt
Méret:
1.94 KB
Formátum:
Item-specific license agreed upon to submission
Leírás: