Discover @hanskrupakar Open Source projects

Hans Krupakar (@hanskrupakar)

hanskrupakar

Stars
201
Global Rank 122,091 (Top 5 %)
Followers 19
Following 2
Registered over 8 years ago
Most used languages

Python
83.3 %

Lua
16.7 %
Location 🇺🇸 United States
Country Total Rank 28,682
Country Ranking

Lua
1,786

Python
5,544

COCO-Style-Dataset-Generator-GUI

A simple GUI-based COCO-style JSON Polygon masks' annotation tool to facilitate quick and efficient crowd-sourced generation of annotation masks and bounding boxes. Optionally, one could choose to use a pretrained Mask RCNN model to come up with initial segmentations.

3D-UNet-pytorch

An implementation of 3D U-Net CNN models for the task of voxel-wise semantic segmentation of 3D MR images for isolation of Low-Grade and High Grade Gliomas, the common types of brain tumour.

Precipitation-Nowcasting

This project uses the Agrimet 15-min Precipitation dataset (2014-2017) to train an RNN prediction network for precipitation using 25 15-min timesteps as input and 6 hour cumulative precipitation as forecast output. The baseline RNN system is then supplemented with the NEXRAD RADAR Images dataset from which the 3-hour precipitation accumulations radar map for 2 years (644 days) and using a CNN-RNN hybrid multimodally tries to predict precipitation.

RNN-In-Numpy

This is a converging bottom up implementation of the Recurrent Neural Network architecture with 3 layers and a tanh activation function using only numpy in Python.

English-Tamil-MT

English Tamil MT is part of a project carried out for ADD (Acoustic Dialect Decoder). It implements Moses based SMT and Neural MT in TensorFlow.

Embedding-And-Morphological-Segmentation

This is part of a research project I am conducting on the effects of morphological segmentation on the embedding process using algorithms like Word2Vec.