ROS node for real-time FCNN depth reconstruction

Concise neural network with C++ and CUDA

Predicting the closing stock price given last N days' data that also includes the output feature for CNN & LSTM, while predicting it for regular NN given only today's data, observing and comparing time series for various models. Additionally finding best value for N previous days and bidirectional LSTM for experiments.

Example of converting CNN network to FCNN using keras.

A collection of Deep Learning labs for the INSAT Data Science course.

This project is about training a deep neural network to identify and track a target in simulation using Udacity's RoboND drone simulator. 🛸 Applications like this are key to many fields of robotics and the techniques applied can be extended to scenarios like advanced cruise control in autonomous vehicles or human-robot collaboration. 👨‍🏫

A ROS1 self-driving car for the "Autonomous Driving Competition". The vehicle itself is a DonkeyCar with RaspberryPi and Raspberry Pi Camera.

This a simple C++ fully connected neural network class for Course COMP3046.

Fully Convolutional Neural Network for heart segmentation

Enjoy the major Deep Learning Projects !!!

Detection of Diabetic Retinopathy using a Fully Connected Neural Network for Data Processing (Tensorflow 1.12.0 and Python 3.6.6)

Classification of 3 species of flowers (versicolor, virginica, setosa) belonging to the Iris family, using a Fully Connected Neural Network for Data Processing (Tensorflow 1.12.0 and Python 3.6.6)

Train FCNN to label each pixel of driving footage as being part of the road (or not). For use in perception systems of autonomous cars.

A simple FCNN and multitheading rProp implementation from scratch in C++.

Implementations of components of Neural Networks from scratch in Numpy

An implementation of the iris flower classification using Keras on the iris dataset

Building fully connected neural network from zero without using deep learning libraries such as Pytorch.

기초딥러닝 Term Project [TO IMPROVE]

Developed a CNN model to classify skin moles as benign or malignant using a balanced dataset from Kaggle, achieving a test accuracy of 81.82% and an AUC of 89.06%. Implemented data preprocessing by resizing images to 224x224 pixels and normalizing pixel values, enhancing model performance and stability.

Ford Otosan Internship Project 2020 || Freespace Segmentation.