FAE CAM: Interactive Neural Style Transfer Exhibit
An interactive real-time video processing application for the Firefly 2023 Arts Festival, applying artistic style transfer and visual effects to a live camera feed.
About This interactive installation allows visitors to see themselves transformed in real-time through various artistic styles. Participants can also choose to capture a stylized picture of themselves.
Details
Real-time video processing using pre-trained neural networks Style transfer based on Logan Engstrom's Fast Style Transfer in TensorFlow Multiple artistic styles available for viewers to cycle through Custom effects including dynamic zooming and smooth transitions between frames
Requirements
Python 3 TensorFlow (>1.0) OpenCV (python3-opencv) PIL/Pillow
Acknowledgements This project builds uses https://github.com/lengstrom/fast-style-transfer
License MIT LICENSE