The ability to obtain graded proportional control is limited by the fundamental challenges with sEMG amplitude resolution and low signal-to-noise Ultrasound imaging provides a non-invasive sensing modality that can spatially resolve individual muscles, including those deep inside the tissue, and detect dynamic activity within different functional compartments in real-time.
Sonomyography-based strategy measures mechanical muscle deformation directly with ultrasound and uses the extracted signals to proportionally control the position of an end-effector Prediction of the type of motion and the completion level from the video sequences of Sonomyography data using Deep Learning Improvement of prediction accuracy at earlier completion levels for better proportional control
There are 9 motions in total and each motion had 5 trials taken from two test subjects and have completion level values from 0-100 I have performed 3 experiments:
Using VGG-16 architecture (Fine Tuned and Off the Shelf)
Using ResNet-34 architecture (Fine Tuned and Off the Shelf)
Using a Off the Shelf ResNet-34 for prediction of motion name with Additional Fully Connected Layers
Data Pre-processing
Converting the Images to 227 × 227 × 3
Dividing the data into train/validation/test sets
Label Encoding
Data Augmentation
Resizing
Center Crop
Horizontal and Vertical Flipping
Rotation
Normalization