dataset-conversion 开源项目推荐（10 个）

ultralytics /JSON2YOLO

用于 COCO、LabelMe、Labelbox、VoTT、INFOLKS 和 ATH 标注的旧版 JSON 转 YOLO 数据集转换器。已被 Ultralytics 包中的 convert_coco() 取代。

arpitg1304 /forge

机器人数据工具包 |在机器人数据集格式（RLDS、LeRobot v2/v3、Zarr、HDF5、Rosbag）之间转换。检查、可视化和分析数据集。可配合HuggingFace Hub使用。专为OpenVLA、Octo、LeRobot和扩散策略工作流打造。

BrenoFariasdaSilva /DDoS-Detector

A Framework for DDoS Attack Detection Using Hyperparameter Optimization, WGAN-GP–Based Data Augmentation, Feature Extraction via Genetic Algorithms, RFE, and PCA, with Individual and Ensemble Classifiers for Multi-Dataset Evaluation.

native-ai /Seg2Yolo

该工具旨在简化将分段标签数据转换为YOLO（你只需看一次）分段格式的过程。

yihong1120 /YOLOv8-Dataset-Transformer

A comprehensive toolkit for converting image classification datasets into object detection datasets and training them using YOLOv8. This project streamlines the process of dataset preparation, augmentation, and training, making it easier to leverage YOLOv8 for custom object detection tasks.

Koooki3 /LEVI

A bilingual LeRobot dataset workbench for visualization, annotation, conversion and quality review.

JaskRendix /pylabelconv

A fast, validated YOLO‑to‑COCO dataset converter with auto‑discovery, strict Pydantic validation, and optional visualization for debugging annotations.

alishavakhi /traffic-sign-light-detection

Single YOLO11 detector recognising 20 traffic sign classes and 13 traffic light states, trained on a unified dataset merged from two independent public datasets.

yashk1103 /Universal-Dataset-Transformation-Pipeline

Comprehensive ML dataset transformation toolkit supporting 15+ formats (SQuAD, CoQA, MS MARCO, Natural Questions) with 3-script pipeline architecture, intelligent conversion algorithms, and batch processing capabilities for large-scale dataset workflows.

A-matindoust /autonomous-driving-object-detection-tracking

YOLO11 based object detection and tracking for autonomous driving, with an option to combine detections with lane perception.