A high-quality neural renderer for visualizing a variety of large scientific point clouds with low motion-to-photon latency.
Exploring the role of data in mitigating racial bias in face recognition systems and also present the flaws in assuming the race-balanced dataset towards fairness.
A novel multi-task learning framework in the absence of common annotated dataset for object detection and saliency estimation in natural images.
Exploration into minimizing the NN model parameters by successive application of knowledge distillation through ancestral- and parental-training.
Augmenting the knowledge distillation framework with sparsity inducing ability of variational inference to learn compact and sparser networks.
A situated visualization method for overlay placement in AR and video applications respecting proximity to the object of interest, non-occlusion, and aesthetics.
A simple ad-hoc algorithm in the image space for visual saliency-aware label placement in AR applications.
Ego-centric gestural interface for AR applications using fingertip regression.
Proposes a method for real-time gesture interaction for AR with egocentric view.