Sparse Autoencoders for Posture Recognition
Proceedings of the International Joint Conference on Neural Networks (IJCNN 2018) ,
pages 2539--2548,
doi: 10.1109/IJCNN.2018.8489726
- Jul 2018
Among different gesture types, static gestures or
postures deliver a broad range of communicative information
like commands or emblems. Vision-based processing for posture
recognition is the most intuitive yet challenging task in intelligent
systems. Achievements in deep learning, specifically convolutional neural networks (CNN), replaced creating hand models
or engineering features for automated image feature learning
at the expense of large data requirements and long training
sessions for optimal parameter tuning. The aim of the present
study is to explore the potentials of sparse autoencoders for
posture recognition, promoting an alternative method to present
convolutional approaches. We conduct experiments with hierarchically designed autoencoders to retain the desired image feature
abstractions on two posture datasets with distinct characteristics.
The different data properties allow us to demonstrate parameter
influences on the network performance. Our evaluation shows
that even a shallow network design achieves superior performance compared to a multiple channel CNN, and comparable
results on a small dataset with sparse image samples. From our
study we conclude that lightweight approaches can be viable
tools for posture recognition, which are worth more explorations
in the future.
@InProceedings{JW18, author = {Jirak, Doreen and Wermter, Stefan}, title = {Sparse Autoencoders for Posture Recognition}, booktitle = {Proceedings of the International Joint Conference on Neural Networks (IJCNN 2018) }, editors = {}, number = {}, volume = {}, pages = {2539--2548}, year = {2018}, month = {Jul}, publisher = {}, doi = {10.1109/IJCNN.2018.8489726}, }