To search this category programmatically: GET https://api.databazaar.io/datasets?category=images
Full API docs: https://api.databazaar.io/llms.txt
Browse / images
Image & Visual Datasets
32 listingsVisual data trains and evaluates computer-vision models — classification, detection, segmentation, and multimodal LLMs. This category includes labeled photo collections, annotated image sets, and video-derived frames with structured metadata. Every listing is scanned for prohibited content before going live, and the free sample lets you inspect annotation quality first-hand.
Shah Ali Azher
7 july 2025 · Updated Oct 26, 2026
This Data is for retouch purpose where people practice for retouching on adobe photoshop
- Format
- JPG
- Downloads
- 0
Sanni Antikainen Raws
7 july 2025 · Updated Oct 26, 2026
This Data is for retouch purpose where people practice for retouching on adobe photoshop
- Downloads
- 0
Conceptual Captions Image–Caption Pairs
Web images paired with natural-language captions harvested from alt-text descriptions, designed for training and evaluating image-captioning and vision-language models.
- Records
- 5,341,263
- Format
- PARQUET
- Downloads
- 2
Vessel Detection Labeled Satellite Image Patches
Validated satellite image patches labeled for maritime vessel detection, with object-detection and classification metadata. Exported from a boat-detection review workflow.
- Records
- 1,920
- Format
- PARQUET
- Downloads
- 2
Global Wheat Full Semantic Organ Segmentation
Labelled images for semantic segmentation of wheat plant organs (canopy, leaves, stems, heads) across diverse field conditions. Dataset spans multiple growing regions and environmental settings.
- Records
- 1,096
- Format
- PARQUET
- Downloads
- 5
RxRx3-core Phenomics Microscopy Image Challenge
Labeled microscopy images of cellular phenotypes from 735 genetic knockouts and 1,674 small-molecule perturbations, designed as a benchmark for learning cellular image representations.
- Records
- 1,335,606
- Format
- PARQUET
- Downloads
- 2
Pexels Synthetic Captions — InternVL2-40B
English captions for Pexels photos generated with InternVL2-40B-AWQ and grounded with original tags. Suitable for text-to-image and image-to-text model training.
- Records
- 567,573
- Format
- PARQUET
- Downloads
- 5
American Sign Language Video Dataset
Video dataset of ASL gestures across 2,208 distinct words, with preprocessed frames suitable for gesture recognition and sign language research.
- Records
- 47
- Format
- PARQUET
- Downloads
- 2
ShotBench Cinematic Understanding Benchmark
Expert-level question-answer pairs evaluating cinematic understanding in vision-language models, drawn from Oscar-nominated cinematography films across images and video clips.
- Records
- 3,572
- Format
- PARQUET
- Downloads
- 2
UAVIT-1M UAV Visual Instruction Tuning Dataset
Instruction-tuning dataset for low-altitude unmanned aerial vehicle visual understanding, covering 11 image- and region-level tasks across 1.24M samples.
- Records
- 1,240,666
- Format
- PARQUET
- Downloads
- 5
TextCaps Image Captioning Benchmark
Image captioning benchmark requiring optical character recognition and text reading within images. Formatted for multimodal model evaluation.
- Records
- 28,408
- Format
- PARQUET
- Downloads
- 4
Vero-600K Multi-Task Visual Reasoning RL Dataset
Reinforcement learning samples spanning 59 datasets across 6 visual reasoning categories, designed for training and evaluating vision-language models.
- Records
- 606,000
- Format
- PARQUET
- Downloads
- 3
PubTabNet Table Images with HTML Annotations
Scientific table images extracted from PubMed Central Open Access articles, paired with HTML structure annotations. Standard benchmark for image-based table recognition and document AI.
- Records
- 223,100
- Format
- PARQUET
- Downloads
- 5
EO-Data-1.5M Interleaved Vision-Text-Action Dataset
Multimodal dataset pairing visual observations, text descriptions, and robot actions to support embodied AI and robot learning tasks. Emphasizes temporal dynamics and causal relationships across modalities.
- Records
- 1,422,808
- Format
- PARQUET
- Downloads
- 2
FLUX-Reason-6M Bilingual Text-to-Image Dataset
Images synthesized by FLUX.1-dev paired with English and Chinese descriptions, designed to develop complex reasoning capabilities in text-to-image generative models.
- Records
- 5,890,279
- Format
- PARQUET
- Downloads
- 1
Text-to-Image Curated Pairs
Text-image pairs in webdataset format designed for fine-tuning text-to-image generative models.
- Records
- 10,000
- Format
- PARQUET
- Downloads
- 3
TreeOfLife-10M Biological Organism Images
Aggregated dataset of biological organism images paired with taxonomic labels across 454,000 taxa, compiled from iNaturalist, BIOSCAN-1M, and Encyclopedia of Life. Used to train vision-language models including BioCLIP.
- Records
- 15,100
- Format
- PARQUET
- Downloads
- 2
MMFineReason-SFT Multimodal Reasoning Samples
123,603 multimodal reasoning samples with Qwen3-VL-235B chain-of-thought traces, curated as the hardest 7% subset where smaller models consistently fail. Includes image and text data in parquet format.
- Records
- 122,603
- Format
- PARQUET
- Downloads
- 4
WorldCuisines VQA Multilingual Visual QA Benchmark
Multilingual visual question-answering dataset on global cuisines spanning 30 languages, with paired images and text questions/answers for vision-language model evaluation and fine-tuning.
- Records
- 1,152,000
- Format
- PARQUET
- Downloads
- 3
TreeOfLife-200M Biodiversity Image Dataset
Biodiversity images spanning 952,257 taxa from GBIF, EOL, BIOSCAN-5M, and FathomNet. Designed for computer vision and zero-shot taxonomic classification tasks.
- Records
- 213,937,319
- Format
- PARQUET
- Downloads
- 4
Nemotron VLM Dataset v2
Vision-language training dataset covering visual question answering, image-text-to-text, video-text-to-text, and document understanding tasks.
- Records
- 4,518,886
- Format
- PARQUET
- Downloads
- 3
FineVision Vision-Language Training Corpus
Image-text dataset with 17.3M images and 88.9M conversational turns. Designed for fine-tuning vision-language models with multi-subset organization.
- Records
- 24,209,105
- Format
- PARQUET
- Downloads
- 2
ScaleEdit-12M Image Editing Dataset
Synthetic instruction–image pairs across 23 editing task families, generated through the ScaleEditor framework. Automated verification may leave residual noise and model biases; instructions are English-only.
- Records
- 12,369,418
- Format
- PARQUET
- Downloads
- 2
Llama-Nemotron Vision-Language Model Dataset
Multimodal dataset for vision-language model training covering visual question answering, optical character recognition, image captioning, and image-to-text tasks. Includes approximately 2.9 million samples.
- Records
- 2,863,834
- Format
- PARQUET
- Downloads
- 2
Other categories