video video 842 1.35k |
|---|
YAML Metadata Warning:The task_categories "video-to-text" is not in the official list: text-classification, token-classification, table-question-answering, question-answering, zero-shot-classification, translation, summarization, feature-extraction, text-generation, fill-mask, sentence-similarity, text-to-speech, text-to-audio, automatic-speech-recognition, audio-to-audio, audio-classification, audio-text-to-text, voice-activity-detection, depth-estimation, image-classification, object-detection, image-segmentation, text-to-image, image-to-text, image-to-image, image-to-video, unconditional-image-generation, video-classification, reinforcement-learning, robotics, tabular-classification, tabular-regression, tabular-to-text, table-to-text, multiple-choice, text-ranking, text-retrieval, time-series-forecasting, text-to-video, image-text-to-text, image-text-to-image, image-text-to-video, visual-question-answering, document-question-answering, zero-shot-image-classification, graph-ml, mask-generation, zero-shot-object-detection, text-to-3d, image-to-3d, image-feature-extraction, video-text-to-text, keypoint-detection, visual-document-retrieval, any-to-any, video-to-video, other
AIC-2026: 15-Video Pilot Dataset & Preprocessing Artifacts
Dataset phục vụ phát triển giao diện (UI) và tính năng truy vấn tìm kiếm (Search / Retrieval) cho cuộc thi AI Challenge 2026 (AIC-2026).
📁 Cấu trúc Dataset
├── videos/ # 15 video MP4 gốc (L21_V001 -> L21_V016)
│ ├── L21_V001.mp4
│ ├── ...
│ └── L21_V016.mp4
├── indexes/ # Vector Index FAISS phục vụ visual search
│ ├── visual-sqfp16.faiss # Index Scalar Quantizer FP16 (41,730 vectors, 81.8MB)
│ ├── visual-sqfp16-manifest.json
│ ├── visual-flat.faiss # Index Flat L2/InnerProduct (163.3MB)
│ └── visual-flat-manifest.json
├── canonical/ # Metadata Store & Identity
│ ├── metadata.sqlite # SQLite database lưu trữ keyframes, shots, transcripts
│ └── dataset-identity.json
├── pilot-15v-canonical.tar.gz # Full archive chứa 42,000+ ảnh keyframes đã trích xuất
└── query_KIS.csv # 53 Benchmark test queries (Known-Item Search)
🚀 Hướng dẫn Sử dụng
1. Tải Artifacts & Video về Local
from huggingface_hub import snapshot_download
snapshot_download(
repo_id="duyaaa/skibidi15",
repo_type="dataset",
local_dir="./data",
)
2. Cấu hình Môi trường cho Backend / API
Đặt các biến môi trường sau trong file .env:
AIC_DATA_DIR="./data"
AIC_V2_ARTIFACTS_DIR="./data/canonical"
AIC_DB_PATH="./data/canonical/metadata.sqlite"
AIC_V2_INDEX_MODE="sqfp16"
AIC_V2_INDEX_PATH="./data/indexes/visual-sqfp16.faiss"
AIC_V2_INDEX_MANIFEST="./data/indexes/visual-sqfp16-manifest.json"
AIC_V2_ENABLE_SERVE="true"
3. Chạy Backend Search API
uvicorn backend.main:app --host 0.0.0.0 --port 8000 --reload
- API Endpoint Tìm kiếm:
POST http://localhost:8000/api/v2/search - Swagger Docs:
http://localhost:8000/docs
4. Đánh giá Benchmark 53 Query KIS
python scripts/run_eval_direct.py
- Recall@1: 56.6%
- Recall@5: 64.2%
- Recall@20: 84.9%
- P50 Latency: ~274 ms
Created for AIC-2026 Pipeline Development.
- Downloads last month
- 60