computer-vision-ai

Python SDK for local and cloud background removal (pip install withoutbg)

Deploy Computer Vision solutions with a few lines of code.

Pytorch implementation of "One-Sided Unsupervised Domain Mapping" NIPS 2017

Computer vision project to automatically recognize digits characters in a seven-segments display

[IROS 2023] Open-Vocabulary Affordance Detection in 3d Point Clouds

State-of-the-art background removal API for images and video. BiRefNet on Modal GPUs, ~200ms per image. 26 endpoints: cutouts, masks, PSD export, upscale, face restore, inpaint, collage, studio shots, video with ProRes 4444 alpha. MIT, self-hostable, 10x cheaper than remove.bg.

C++ inference implementation of PaddleOCR

MouthMind is a multi-view, sequence-to-sequence framework for visual speech recognition. It directly translates video frames into character sequences, without the need for autoregressive teacher forcing, resulting in robust, emergent learning of spoken words.

A website integrating an AI model to convert sign language pictures and videos to text. Has a more than mAP50-95 of more than 90% and a response time of less than 30ms.

A computer vision project for detecting student concentration levels using image classification techniques. The concentration levels can be one amongst the following two: engaged and not engaged.

[ConvLSTM] - Simulation of drone navigation through a procedurally generated disaster environment in CoppeliaSim, using a quadcopter equipped with an RGBD camera. Includes manual keyboard control, waypoint navigation, and dynamic obstacle generation.

The main abnormal behaviors that this project can detect are: Violence, covering camera, Choking, lying down, Running, Motion in restricted areas. It provides much flexibility by allowing users to choose the abnormal behaviors they want to be detected and keeps track of every abnormal event to be reviewed. We used three methods to detect abnorma…

Cross-platform SAM-assisted image annotation and resumable expert review with polygon editing, audit trails, change detection, timing, provenance, and CSV export.

RoadGuard AI is an advanced, automated road infrastructure monitoring and civic pothole complaint reporting system. It integrates computer vision (YOLOv8 Segmentation), geolocational intelligence, dynamic web scraping, and official communication channels into a unified road safety pipeline.

AmazingScanner is a document scanner built from scratch. It detects page corners, rectifies tilted photos with homography, removes shadows and lighting artifacts, enhances degraded pages to scanner quality, and produces upright, styled, exportable scans. Includes corner detection, OCR evaluation, synthetic degradations, and an end-to-end scanner.

Touchless music control via hand gestures — CNN + OpenCV, no keyboard needed. Real-time hand gesture recognition to play/pause and control volume, built with PyTorch & OpenCV.

Machine Learning and Deep Learning Courses on YouTube

Image and video detector. Program can scan all images and videos in folder using AI model and select those that match specified target. Processing is done locally.

Counterfeit product detector for marketplaces: parses WB / Ozon / Yandex Market listings and aggregates verdicts from multiple vision models with batch processing