Advanced lane detection (incl. curvature) using advanced computer vision techniques.

A pipeline to detect road lane lines in images from a monocular camera

This tool can be used to find the most influential words on a document. We define most influential as the words that influence a trained classifier the most to give it a particular classification.

Object detection and tracking in videos using HOG+SVM and Kalman Filter uses Filterpy,Numpy and Sci-kit learn

tf.image.resize_images has aliasing when downsampling and does not have gradients for bicubic mode. This implementation fixes those problems.

Neural network visualization tool after an optional model compression with parameter pruning: (integrated) gradients, guided/visual backpropagation, activation maps for the cao model on the IndianPines dataset

Master thesis for the MSc. Artificial Intelligence at the Universiteit van Amsterdam, 2019

First assignment in ׳Deep Learning for Texts and Sequences' course (using NumPy only) by Prof. Yoav Goldberg at Bar-Ilan University

Advanced Lane Finding (project 2 of 9 from Udacity Self-Driving Car Engineer Nanodegree)

5533 Linear Gradients free to use. Includes a script to generate RGB gradients from YUV colors

Demonstrate how to do backpropagation using an example of BatchNorm-Sigmoid-MSELoss network with a detailed derivation of gradients and custom implementations.

Software pipeline to identify lane boundaries from a video streaming from a front-facing camera on a car using color transform and gradient

Explore the Math behind it by designing a neural network, derive the parameter gradients with respect to loss function and update the parameter weights and update the weight parameters using the gradients without the help of in-built libraries.

Predicting wine quality using regression on the well-known UCI data set and more

Python script to randomly generate gradient artworks

An automatic differentiation library written in Python with NumPy vectorization.

Python Library for creating and training CNNs. Implemented from scratch.

Avoiding the vanishing gradients problem by adding random noise and batch normalization