Expedition Technology

Researched novel semi-supervised framework for detecting out-of-distribution inputs. Implemented the OpenMax algorithm: a methodology to adapt deep networks for open set recognition - estimates the probability of an input being from an unknown class in Tensorflow. Investigated existing literature in the domain of Open Set Recognition and performed A-B testing.

Researched anchorless object detection techniques for 3D point cloud object detection. Designed a convolutional neural network on the basis of existing VoxelNet and CenterNet architectures in Tensorflow.