coco-format

✏️ Web-based image segmentation tool for object detection, localization, and keypoints

Cityscapes to CoCo Format Conversion Tool for Mask-RCNN and Detectron

Utility scripts for COCO json annotation format

Fast alternative to FiftyOne for creating a subset of the COCO dataset.

Tool for working with images annotations. Allows you to convert, modify and analyze annotations to images of such formats as Yolo, COCO, LabelMe, etc.

🔄 Converts horizontal bounding boxes into oriented (rotated) boxes by prompting SAM-family models. Includes model ensembling, confidence scoring, evaluation, format conversion, and hyperparameter tuning.

Convert annotations from VIA to COCO.

Vehicle make and model recognition (VMMR) using BVMMR dataset v2

A dataset for semantic segmentation of Sosnowsky's hogweed in the ground-level view photos taken in St. Petersburg, Malaya Vishera, Pushkin, etc.

Dataset prep tool for object detection tasks

mask rcnn training with coco-like dataset. You can use for trainnig your own coco.json (polygon) dataset in Google Colab.

Object Detection Tools to convert between dataset formats, image splitting and stitching.

VisionForge is an offline-first AI-assisted annotation, segmentation, and dataset intelligence tool for Edge AI, ADAS, and computer vision datasets.

Reusable toolkit for generating wide field-of-view, hemispherical and fisheye object-detection datasets from standard perspective images.

Real-time object detection & tracking pipeline — Mask R-CNN + SORT algorithm with Kalman filtering. 78.4% tracking accuracy, 100% ID stability. Self-supervised evaluation metrics.

SnapSeg: Fast interactive segmentation annotation tool with SAM, point prompts, and COCO/YOLO export.

HalalBench: A Multilingual OCR Benchmark for Food Packaging Ingredient Extraction — halal food app, halal scanner

Standalone pip-installable bidirectional COCO↔YOLO dataset format converter with multiprocessing support

Train RTMPose on custom keypoint datasets. Includes YOLO-to-COCO preprocessing, 18-keypoint config, and full training/inference workflow for MMPose.

Deep Learning project for waste object detection and segmentation using YOLOv8 and U-Net on the TACO dataset. Implements object detection with YOLOv8 for bounding boxes and custom U-Net for pixel-wise segmentation. Includes EDA, data augmentation pipeline, and comparative analysis. Built with PyTorch for CS4045 course project.