ucuapps Goto Github PK

brats2021_challenge

computer-vision-course

Computer Vision course for CS bachelors in UCU (2019)

coronaryarterystenosisscoreclassification

CNN for Classification of Coronary Artery Stenosis Score inMPR Images.

dino-medical

PyTorch code for Vision Transformers training with the Self-Supervised learning method DINO

inkscapebarreldistortion

An Inkscape extension for lens distortion of objects

kornia

Open Source Differentiable Computer Vision Library

lidchallenge2020-nopeopleallowed

A 3rd place solution for LID Challenge at CVPR 2020 on Weakly Supervised Semantic Segmentation

modelicagym

Modelica models integration with Open AI Gym

modified-maskformer-for-polyps-segmentation

Mask Classification-based method for Polyps Segmentation and Detection (EndoCV 2022 challenge)

openglue

Open Source Graph Neural Net Based Pipeline for Image Matching

portfolio-apps

Portfolio for APPS students

robert

Mobile robot

robert_the_robot

This repository is devoted to the development of the hardware implementation of the Robert.

robust-dl-pipeline-for-pvc-localization

Premature ventricular contraction(PVC) is among the most frequently occurring types of arrhythmias. Along with other cardiovascular diseases, it may easily cause hazardous health conditions, making PVC detection task extremely important in cardiac care. However, the long-term nature of monitoring, sophisticated morphological features, and patient variability makes the manual observation of PVC an impractical task. Existing approaches for automated PVC identification suffer from a range of disadvantages. These include domain-specific handcrafted features, usage of manually delineated R peaks locations, tested on a tiny sample of PVC beats(usually a small subset of MIT-BIH database). We address some of these drawbacks in proposed framework, which takes a raw ECG signal as an input and localizes R peaks of the PVC beats. It consists of two neural networks. The first one is an encoder-decoder architecture that localizes the R peak of both Normal and anomalous heartbeats. Provided R peaks positions, our CardioIncNet model, adopted for ECG signal data, does the delineation of healthy versus PVC bits. We have performed the extensive evaluation of our pipeline with both single- and cross-dataset paradigms on three public datasets. Our approach results in over 0.99 and 0.979 F1-measure on both single- and cross-dataset paradigms for R peaks localization task and above 0.96 and 0.85 F1 score for the PVC beats classification task.

single-view-autocalib

Single-View Auto-Calibration from Lens-Distorted Images of Urban Scenes. In WACV (2021)

synthtext

Code for generating synthetic text images as described in "Synthetic Data for Text Localisation in Natural Images", Ankush Gupta, Andrea Vedaldi, Andrew Zisserman, CVPR 2016.

top-view-multi-person-tracking

This repo contains links to multi-person re-identification and tracking dataset in top view multi-camera environment.

ukrainian-vwsd-benchmark

wsmis

Weakly Supervised Medical Images Segmentation