Sharpen Your AI with Better Data

Nexdata Trending datasets

Gain access to off-the-shelf, up-to-date datasets to drive your business growth.

10,000-Hour Egocentric Video Dataset for Robotics and AI Manipulation Training

This dataset contains 10,000 hours of egocentric multimodal data collected from diverse real-world environments, including residential, retail, and office scenarios. It covers a wide range of human activities and manipulation tasks, such as meal preparation, cleaning, storage, garment care, merchandising, and object picking. Each sample includes synchronized 4K stereo video, camera calibration parameters, 76-point full-body pose annotations, and fine-grained step-by-step action sequence labels. The dataset is suitable for robot learning, manipulation policy development, and Vision-Language-Action (VLA) models.

embodied ai dataset robotics dataset robot learning dataset robot manipulation dataset vla dataset egocentric video dataset

1,042 Segments 6-camera Egocentric Embodied AI Dataset

This dataset contains 1,042 egocentric video segments (approximately 35 seconds each) collected across 34 locations and 6 real-world environments, including homes, offices, and retail scenarios. Powered by self-developed VSLAM system, achieving millimeter-level positioning, multi-sensor hard-triggered synchronization (≤1ms), and a high frame rate of 60fps for RGB. It includes multi-view videos, calibrations, point clouds, SLAM trajectories, and gesture recognition results in standard formats. Designed for embodied AI, spatial perception, and 3D reconstruction, it offers high precision, diverse scenarios, and out-of-the-box usability, making it ideal for training robust perception-action models.

embodied ai dataset robot learning dataset robotics training data egocentric dataset robot perception dataset multimodal robotics dataset SLAM dataset

Agent Trajectory Dataset for Tool-Use Training and AI Agent Evaluation

This dataset covers office-based scenarios such as in-depth searches, data analysis, and industry research, encompassing complete multi-turn reasoning trajectories and tool-calling chains. It is designed to support the analysis of agent planning capabilities, research into tool selection strategies, and quality assessment, providing a structured benchmark for agent training and evaluation.

ai agent dataset agent training dataset agent trajectory dataset llm agent dataset tool use dataset agent evaluation datase

1.51M Instruction-Based Image Editing Dataset for Generative AI Training

This dataset contains 1.51 million annotated image editing pairs. Editing types include 500,000 sets of portrait/object consistency editing, 300,000 sets of structural edits, 210,000 sets of mixed editing, and 450,000 sets of spatial editing, and 50,000 sets of style transfer editing. The editing targets cover scenes such as people, animals, goods, plants, and landscapes. In terms of annotation, the targets that need to be edited in the image are edited according to the editing instructions. The data can be used for tasks such as image synthesis, data augmentation, and virtual scene generation.

generative AI image dataset image editing dataset AI image editing dataset image editing training data AI image manipulation dataset image editing pairs dataset image inpainting dataset style transfer dataset

50,000 Image Editing Datasets – Object Removal, Addition & Modification Dataset for AI Training

50,000 Sets - Image Editing Data. The editing types include human attribute editing, image semantic editing, and image structure editing. The editing targets cover scenes such as people, animals, goods, plants, and landscapes. In terms of annotation, based on the editing instructions, the targets that need to be edited in the image are edited. The data can be used for tasks such as image synthesis, data augmentation, and virtual scene generation.

image editing dataset image synthesis data object removal dataset object addition data AI image generation dataset virtual scene dataset annotated image editing data inpainting dataset AI training data for image manipulation generative image dataset

205 Hours - Japanese Full-Duplex Multi-Channel Speech Dataset

205 Hours Japanese Full-Duplex Multi-Channel Speech Dataset is collected from dialogues based on given topics. Transcribed with text content, speaker's ID, gender, age and other attributes. Our dataset was collected from extensive and diversify speakers, geographicly speaking, enhancing model performance in real and complex tasks. Quality tested by various AI companies. We strictly adhere to data protection regulations and privacy standards, ensuring the maintenance of user privacy and legal rights throughout the data collection, storage, and usage processes, our datasets are all GDPR, CCPA, PIPL complied.

Japanese speech dataset spontaneous Japanese dialogue multi-stream Japanese audio data conversational Japanese corpus Japanese voice dataset full-duplex speech dataset multi-stream speech dataset multi-channel audio dataset

200 Hours Korean Full-Duplex Multi-Channel Speech Dataset

This 200 Hours Korean Full-Duplex Multi-Channel Speech Dataset features multi-stream audio recorded via smartphones, simulating natural conversations across a range of everyday topics. Each dialogue is annotated with transcripts, speaker ID, gender, and age. Collected from diverse speakers across various regions in Korea, the dataset enhances AI model robustness in real-world applications. Ideal for training automatic speech recognition (ASR), conversational AI, multilingual dialogue systems, and natural speech processing models. All data complies with GDPR, CCPA, and PIPL privacy standards, ensuring safe and ethical AI training.

Korean speech dataset spontaneous dialogue Korean multi-stream audio dataset conversational Korean speech smartphone-recorded audio dual-speaker dataset real-world Korean conversation full-duplex speech dataset

172 Hours - American English Full-Duplex Multi-Channel Speech Dataset

172 Hours - American English Full-Duplex Multi-Channel Speech Dataset, collected from dialogues based on given topics. Transcribed with text content, speaker's ID, gender, age and other attributes. Our dataset was collected from extensive and diversify speakers, geographicly speaking, enhancing model performance in real and complex tasks. Quality tested by various AI companies. We strictly adhere to data protection regulations and privacy standards, ensuring the maintenance of user privacy and legal rights throughout the data collection, storage, and usage processes, our datasets are all GDPR, CCPA, PIPL complied.

American English speech dataset multi-stream speech dataset full-duplex dialogue dataset spontaneous speech dataset smartphone speech data multi-channel audio dataset speech recognition training data dialogue AI dataset

Long Context Reasoning Dataset – Multi-Language (EN/CH/KR) Benchmark for LLM Evaluation

This dataset is designed to tackle the core weaknesses of today's large language models when it comes to processing long documents and performing complex reasoning. It consists of 7,500 high-quality training examples across three languages—Chinese, English, and Korean. Each instance is built around a long-text passage and includes questions that require synthesizing information across paragraphs and documents, while following multi-step logical chains. The goal is to offer a thorough and rigorous evaluation framework that tests a model's ability to perceive long-range context, retrieve relevant information, construct sound reasoning paths, and trace evidence back to its source.

long context dataset long context reasoning dataset LLM long context dataset long document QA dataset multi hop reasoning dataset reasoning dataset for LLM multi step reasoning dataset

116,048 Sets – 3D Hand Pose & Gesture Recognition Dataset

This dataset contains 116,048 sets of 3D handpose data, each set includes hand mask image(RGB, 24-bit), depth image(16-bit), camera intrinsic parameter file(TXT), 3D keypoints file(OBJ), mesh file(OBJ), gesture type file(TXT), keypoints demo image(JPG), and mesh demo image(JPG). The data is collected indoors, with the right hand (no handheld objects), covering both first-person and third-person perspectives, multiple gesture types, finger poses, hand overall rotation poses, individuals and Kinect devices used. This dataset does not include personally identifiable facial information, with hand mask images and depth images aligned. This dataset can be used for tasks such as handpose recognition, hand 3D reconstruction, and hand keypoints detection.

3D hand pose dataset hand keypoints dataset hand gesture recognition RGB-D hand dataset hand mesh dataset AI training hand pose 3D hand reconstruction computer vision hand dataset gesture recognition data

FAQ

How does Nexdata ensure the quality and accuracy of its AI training datasets?

Nexdata uses a multi-stage quality control process covering data collection, annotation, validation, and final delivery. Depending on the project, we can provide customized quality standards, multi-level annotation review, consistency checks, and sample-based quality inspection to ensure the dataset meets specific model training requirements.

Can Nexdata customize datasets according to our AI model and project requirements?

Yes. Nexdata provides customized data collection, annotation, and curation services based on your model type, application scenario, target languages, data volume, annotation format, and quality requirements. We can support projects ranging from computer vision and autonomous driving to speech, OCR, LLM, VLM/VLA, and embodied AI.

What are the licensing and commercial usage options for Nexdata datasets?

Nexdata provides commercially licensed AI training datasets for enterprise use. Licensing terms can be tailored according to the intended application, geographic scope, usage scenario, and project requirements. Customers can also discuss specific licensing requirements with our team before purchasing a dataset.