frontend/src/data/datasets.ts - master - FlysAt/DataFoundry

frontend/src/data/datasets.ts
121 строка4 KB
FlysAt8
front update
25 мар 2026, 19:52
25 мар 2026, 19:520976797
import type { Dataset } from '../types';

export const datasets: Dataset[] = [
  {
    id: 1,
    name: "ImageNet-1K Pro",
    category: "computer-vision",
    description: "Enhanced ImageNet with 1.28M training images, improved annotations, and balanced classes for robust image classification.",
    size: "150 GB",
    samples: "1,281,167",
    price: 299,
    rating: 4.9,
    reviews: 2847,
    tags: ["Image Classification", "ResNet", "CNN"],
    image: "http://static.photos/technology/640x360/201"
  },
  {
    id: 2,
    name: "Common Crawl NLP",
    category: "nlp",
    description: "Cleaned and deduplicated web crawl data, tokenized and ready for language model training. Multiple languages supported.",
    size: "2.8 TB",
    samples: "40B tokens",
    price: 599,
    rating: 4.8,
    reviews: 1523,
    tags: ["LLM Training", "Transformer", "BERT"],
    image: "http://static.photos/technology/640x360/202"
  },
  {
    id: 3,
    name: "LibriSpeech ASR",
    category: "audio",
    description: "High-quality audiobook recordings with aligned transcripts. Perfect for automatic speech recognition and TTS research.",
    size: "60 GB",
    samples: "1,000 hrs",
    price: 149,
    rating: 4.7,
    reviews: 892,
    tags: ["ASR", "Speech", "Whisper"],
    image: "http://static.photos/technology/640x360/203"
  },
  {
    id: 4,
    name: "Tabular Bank Fraud",
    category: "tabular",
    description: "Anonymized banking transactions with fraud labels. Includes temporal features and customer behavioral patterns.",
    size: "12 GB",
    samples: "6.3M rows",
    price: 399,
    rating: 4.9,
    reviews: 634,
    tags: ["Fraud Detection", "XGBoost", "Finance"],
    image: "http://static.photos/finance/640x360/204"
  },
  {
    id: 5,
    name: "Crypto Price History",
    category: "time-series",
    description: "Minute-level OHLCV data for top 100 cryptocurrencies. Includes on-chain metrics and sentiment indicators.",
    size: "45 GB",
    samples: "500M points",
    price: 199,
    rating: 4.6,
    reviews: 445,
    tags: ["Forecasting", "LSTM", "Trading"],
    image: "http://static.photos/finance/640x360/205"
  },
  {
    id: 6,
    name: "COCO 2024 Extended",
    category: "computer-vision",
    description: "Object detection and segmentation dataset with 330K images, 2.5M object instances, and 5 captions per image.",
    size: "80 GB",
    samples: "330K images",
    price: 249,
    rating: 4.8,
    reviews: 2103,
    tags: ["Object Detection", "Segmentation", "YOLO"],
    image: "http://static.photos/technology/640x360/206"
  },
  {
    id: 7,
    name: "Medical Imaging CT",
    category: "computer-vision",
    description: "3D CT scans with radiologist annotations for tumor detection. Includes DICOM metadata and segmentation masks.",
    size: "500 GB",
    samples: "50K scans",
    price: 899,
    rating: 5.0,
    reviews: 312,
    tags: ["Medical AI", "3D CNN", "Healthcare"],
    image: "http://static.photos/medical/640x360/207"
  },
  {
    id: 8,
    name: "GitHub Code Corpus",
    category: "nlp",
    description: "Curated code repositories across 50+ languages. Includes documentation, commit messages, and code reviews.",
    size: "1.2 TB",
    samples: "100M files",
    price: 449,
    rating: 4.7,
    reviews: 876,
    tags: ["Code Generation", "Copilot", "AST"],
    image: "http://static.photos/technology/640x360/208"
  },
  {
    id: 9,
    name: "Autonomous Driving",
    category: "computer-vision",
    description: "Multi-sensor dataset with LiDAR, camera, and radar data. Includes 3D bounding boxes and trajectory predictions.",
    size: "2.5 TB",
    samples: "1,000 hrs",
    price: 1299,
    rating: 4.9,
    reviews: 523,
    tags: ["Autonomous", "LiDAR", "Sensor Fusion"],
    image: "http://static.photos/automotive/640x360/209"
  }
];