Research Engineer Intern

Neurala

May 2018 - December 2018
Internship_Report


Internship Timeline

Implemented transfer learning, pre-training, training neural networks, device testing (using Android Studio) and crafting new architectures for tasks including video segmentation, semantic/instance segmentation, object detection, and classification.

Designed and debugged a unified training pipeline in Python for TensorFlow (TF Slim), Keras, MXNet, and Caffe frameworks.

Deployed semantic segmentation model in specialized embedded hardware with compression techniques as well as preparing model by optimizing neural networks to run efficiently on edge device for the World’s leading non-US based mobile company.

Improved and benchmarked a novel one-shot-learning method for object detection and segmentation using python, GluonCV.

Generated nearly 200k images with synthetic data generation technique using MATLAB and Brainbuilder image tagging tool.

Improved the quality of image and video processing for scientific development using computer vision techniques of customer facing technology demos and contractual deliveries.


Reviews