Discover AI datasets
across platforms
Recure helps machine learning teams discover and evaluate datasets across sources with AI-powered search and standardized metadata.
AI-Powered Aggregation
Our agent continuously scans connected sources, surfacing new and updated datasets automatically.
dataset for self-driving
NLM
Describe what you need in plain language and our semantic engine handles the rest.
Cityscapes Semantic
25,000 images · 50 cities
High-quality pixel level annotations for street scenes.
segmentation
autonomous-driving
urban
pixel-level
● 42
50.0k
1000
Each dataset is standardized with tags, summaries, task and modality classification, and a directional health score.
Cityscapes
BDD100K
Compare datasets on classes, modalities, license, file types, and quality signals in one unified view.
Trending Now
Most viewed datasets this week
Browse by Category
Explore datasets organized by domain and use case
Recently Added
Newly discovered datasets from the past week
Featured Datasets
Handpicked high-quality datasets from across the web
Benchmark Datasets
Standard datasets with publicly available baseline metrics and leaderboards
COCO (Common Objects in Context) - Object Detection
Baseline Metrics
accuracy
89.2
map
0.4
iou
0.87
fOneScore
0.86
COCO-Seg (2026 Edition)
Baseline Metrics
accuracy
85.2
map
33.9
iou
0.87
fOneScore
0.86
Open Images V7+ (Google)
Baseline Metrics
accuracy
88.2
map
0.76
iou
0.87
fOneScore
0.86
Cityscapes Dataset
Baseline Metrics
miou
86.2
accuracy
91.5
fOneScore
0.88
ADE20K
Baseline Metrics
miou
57.8
accuracy
82.1
pixelAccuracy
81.4
COCO Panoptic
Baseline Metrics
pq
57.8
sq
83.2
rq
68.4
MMLU-Pro
Baseline Metrics
accuracy
72.6
fOneScore
0.71
cot boost
12.5
HumanEval+
Baseline Metrics
pass at 1
86.2
pass at 10
94.5
edge case fail rate
4.2
GSM8K
Baseline Metrics
accuracy
96.4
reasoning steps avg
4.5
error rate
2.1
Academic Datasets
Recently released datasets from research papers with EDA and analytics
Infinity-Chat
26k+ open-ended prompts designed to expose 'mode collapse' in LLM creativity.
Analytics Summary
WorldModelBench
350+ physics-constrained scenarios to test video generation logic.
Analytics Summary
LiveBench
Monthly updated questions from arXiv, math competitions, and news.
Analytics Summary
STSBench (STSnu)
43 diverse driving scenarios with 971 verified spatial reasoning QA pairs.
Analytics Summary
WildBench
Real-world user queries evaluated using automated pairwise comparison.
Analytics Summary
OS-Marathon
Benchmarks agents on long-horizon, repetitive desktop tasks to test stability.
Analytics Summary
XLRS-Bench
Ultra-high resolution (8.5k x 8.5k) remote sensing for MLLMs.
Analytics Summary
MuVR
Retrieving specific moments from long, untrimmed video streams.
Analytics Summary
AODRaw
7,700+ RAW sensor images for detection in adverse weather.
Analytics Summary
Request Datasets
Didn't find what you were looking for? Submit a request and we'll help you find or create the dataset you need.
Request dataset