Multimodal Egocentric Vision -- 2
Thanh toán khi bàn giao
The primary goal of this project is to explore egocentric vision and the application of multimodal learning, with a specific focus on incorporating ElectroMyoGraphy (EMG) data. The project is divided into three main steps. Firstly, a thorough literature review is conducted to gain familiarity with egocentric vision and multimodal learning techniques. Secondly, existing pre-trained models for action recognition are utilized to extract features from RGB streams and train a classifier for action recognition tasks. Finally, the project investigates the integration of EMG data by either designing a specialized model for gesture recognition using the Myo Armband or developing a model to reconstruct EMG data from an RGB stream. The overall aim is to gain insights into egocentric vision and explore the potential of incorporating EMG data for improved action recognition.
(the datasets will be given to you after negotiation to doing it)
ID dự án: #36653483