elsa9421 Goto Github PK

cartpole

Reinforcement learning for CartPole problem using the OpenAI toolkit.

checkpoints

Checkpoints for various trained algorithms

computer-vision-hw1-pet-edge-detection

computer-vision-image-blending

Image blending Image pyramids are image representations useful for many downstream applications. This problem uses pyramid image processing to blend two images.

computer-vision-image-transalation

Implement pix2pix In this problem set, we will implement an image-to-image translation method, based on pix2pix [1]. We hope that this will give you experience reading a modern computer vision paper and implementing the architecture described within it. We’ll train a model to generate pictures of facades from label maps, using the CMP Facade Database [2]. This dataset contains 606 rectified images of facades from various sources, which have been manually annotated. Below is a sample image from the database. In this homework, we will implement GAN to translate labels into facade images.

computer-vision-motion-magnification-and-texture-synthesis

Motion Magnification and Texture Synthesis

computer-vision-object-recognition

In this problem set, we will implement a single-stage object detector, based on YOLO v1 [1] and v2 [2]. Unlike the (better-performing) R-CNN models, single-stage stage detectors predict bounding boxes and classes without explicitly cropping region proposals out of the image or feature map. This makes them significantly faster to run. We’ll train and evaluate our detector on the PASCAL VOC dataset, a standard dataset for object detection tasks. The full dataset contains a total of 11K train/val data images with 27K labeled objects, spanning 20 classes (Figure 1). To make training faster, though, we will only use a subset that contains 2.5K images.

computer-vision-representation-learning

A simple self-supervised learning method: an autoencoder.

computer-vision-scene-recognition

In this problem set, you will train a CNN to solve the scene recognition problem, i.e., the problem of determining which scene category a picture depicts. You will train two neural networks, which we call MiniVGG and MiniVGG-BN. MiniVGG is a smaller, simplified version of the VGG [1] architecture, while MiniVGG-BN is identical to MiniVGG except that we added batch normalization layers after each convolution layer. You will train the neural networks on the MiniPlaces dataset1 . We’ll use 90,000 images for training, 10,000 images for validation, and the remaining 10,000 images for testing. Sample images from MiniPlaces dataset (along with their categories) a

cyclegan

https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix

cyclegan_readable

deep-learning-for-vision

Colab and .py files developed for EECS 598 - Deep Learning for Vision.

deepdesign2019

Welcome to the code repository for ARCH 660/662, the Deep Design Studio! Please Scroll down to the bottom of the page for how to use the code in this repository

elsa9421-computer-vision-object-detection

In this problem set, we will implement an image-to-image translation method, based on pix2pix [1]. We hope that this will give you experience reading a modern computer vision paper and implementing the architecture described within it. We’ll train a model to generate pictures of facades from label maps, using the CMP Facade Database [2]. This dataset contains 606 rectified images of facades from various sources, which have been manually annotated. Below is a sample image from the database. In this homework, we will implement GAN to translate labels into facade images.

gmm_live_plotting

Code that produces a video showing convergence of the EM algorithm for GMMs on a simple 2d dataset

independent-component-analysis

maximum-likelihood Independent Component Analysis (ICA) for blind audio separation

interactive-ipython-demos

This repository includes a collection of IPython Notebooks developed from June 2020 through August 2020. These notebooks were developed by me for the Machine Learning Course EECS 545 taught by Prof. Clayton Scott at UMich for Fall 2020. Any references used are inlcluded in the notebooks. Most of the IPython demos (Google Colab) are interactive, enabling Data Visualization for Machine Learning algorithms.

it-cert-automation-practice

Google IT Automation with Python Professional Certificate - Practice files

jigsawnet

JigsawNet: Shredded Images Reassembly Using Convolutional Neural Network and Loop-based Composition

knn_demo

Demonstration of K- Nearest Neighbors algorithm, from scratch as well as using predefined python libraries. Using Banana Dataset, and plotting output with k as a variable parameter. Accuracy calculated using confusion matrix. Includes plot of Error rate vs k

linear-discriminant-analysis

Linear Discriminant Analysis: The code creates a plot showing two Gaussian classes in 2d. They are parameterize the covariance matrix by parameters lambda_1 and theta, and has an interactive widget which shows the LDA hyperplane as the parameter varies. It also plots elliptical contours of the Gaussian densities and a small number of data points. The parameters should control the eccentricity of the ellipse associated to the covariance matrix.

machine_learning_cnn_rnn_transfe_learning

machine_learning_kmeans_gmm_imagecompression

K-Means Clustering and GMM for Image Compression

offadedbluequilts

An on going writing experiment. Nothing fancy here.

optimal-separating-hyperplanes

Demo1 showing the different hard margin hyperplanes possible

pca

Apply PCA to face images

problem-set-4-backpropagation

Backpropagaton

project_504_cyclegan_sara_elsa

python_wheel_of_fortune

Wheel_of_Fortune_code

rnn_application

Checkpoints for different models