I am interested in problems at the intersection of natural language processing and computer vision.
Generating Animated Videos of Human Activities from Natural Language Descriptions
Angela S. Lin*,
Raymond J. Mooney
Visually Grounded Interaction and Language Workshop at NeurIPS, 2018
A sequence-to-sequence model for 3D skeletal human generation from text where the animation decoder is
pretrained using an autoencoder.
Note: * denotes co-first authorship
Real-Time User-Guided Image Colorization with Learned Deep Priors
Angela S. Lin,
Alexei A. Efros.
A network trained using simulated user interaction from ground truth colors produces surprisingly good colorization results with real human users.