19,997 machine learning datasets
19,997 dataset results
Video-ChatGPT introduces the VideoInstruct100K dataset, which employs a semi-automatic annotation pipeline to generate 75K instruction-tuning QA pairs. To address the limitations of this annotation process, we present \ourdata~dataset developed through an improved annotation pipeline. Our approach improves the accuracy and quality of instruction tuning pairs by improving keyframe extraction, leveraging SoTA large multimodal models (LMMs) for detailed descriptions, and refining the instruction generation strategy.
We present the World Wide Dishes dataset which seeks to assess disparities in representations of food through a decentralised data collection effort to gather perspectives directly from people with a wide variety of backgrounds from around the globe with the aim of creating a dataset consisting of their insights into their own experiences of foods relevant to their cultural, regional, national, or ethnic lives.
This task stems from the observation that text embedded in images is intrinsically different from common visual elements and natural language due to the need to align the modalities of vision, text, and text embedded in images.
Foley audio, critical for enhancing the immersive experience in multimedia content, faces significant challenges in the AI-generated content (AIGC) landscape. Despite advancements in AIGC technologies for text and image generation, the foley audio dubbing remains rudimentary due to difficulties in cross-modal scene matching and content correlation. Current text-to-audio technology, which relies on detailed and acoustically relevant textual descriptions, falls short in practical video dubbing applications. Existing datasets like AudioSet, AudioCaps, Clotho, Sound-of-Story, and WavCaps do not fully meet the requirements for real-world foley audio dubbing task. To address this, we introduce the Multi-modal Image and Narrative Text Dubbing Dataset (MINT), designed to enhance mainstream dubbing tasks such as literary story audiobooks dubbing, image/silent video dubbing. Besides, to address the limitations of existing TTA technology in understanding and planning complex prompts, a Foley Audi
Click to add a brief description of the dataset (Markdown and LaTeX enabled).
The COCO-WAN benchmark is designed to assess the impact of weakly annotations (combined with auto-annotation tools) noise on instance segmentation models. This benchmark is built upon the COCO dataset and incorporates noise generated through weak annotations, simulating real-world scenarios where annotations might be imperfect due to semi-automated tools. It includes various levels of noise to challenge the robustness and generalization capabilities of segmentation models.
PolyNews is a multilingual dataset containing news titles in 77 languages and 19 scripts.
PolyNews is a multilingual parallel dataset containing news titles 833 language pairs, spanning in 64 languages and 17 scripts.
This folder contains 5-arcmin resolution maps for the application rate of each fertilizer (N,P2O5,K2O) on cropland and each of the 13 major crop groups resulting from our study.
OffensiveLang is a community based implicit offensive language dataset generated by ChatGPT 3.5 containing data for 38 different target groups. It has been meticulously annotated by Amazon MTurk workers, ensuring high-quality labeling of hate speech. Additionally, a prompt-based zero-shot method was employed with ChatGPT and the detection results were compared between human annotation and ChatGPT annotation. This dataset is invaluable for researchers and practitioners working on implicit hate speech detection and large language models.
HateSpeechCorpus is a dataset collected from Twitter, consisting of 3003 tweets. It has been meticulously annotated by three speech-language pathology graduate students, ensuring high-quality labeling of hate speech. This dataset is invaluable for researchers and practitioners working on hate speech detection and natural language processing.
BenBench is designed to benchmark the potential for data leakage in benchmark datasets, which can lead to biased and inequitable comparisons. In this work, we are not pursuing technical contributions in system development; instead, we are attempting to encourage the healthy development of this field, particularly through the lens of mathematical reasoning tasks, in the following aspects:
Spoken Named Entity Recognition (NER) aims to extracting named entities from speech and categorizing them into types like person, location, organization, etc. In this work, we present VietMed-NER - the first spoken NER dataset in the medical domain. To our best knowledge, our real-world dataset is the largest spoken NER dataset in the world in terms of the number of entity types, featuring 18 distinct types. Secondly, we present baseline results using various state-of-the-art pre-trained models: encoder-only and sequence-to-sequence. We found that pre-trained multilingual models XLM-R outperformed all monolingual models on both reference text and ASR output. Also in general, encoders perform better than sequence-to-sequence models for the NER task. By simply translating, the transcript is applicable not just to Vietnamese but to other languages as well. All code, data and models are made publicly available here: https://github.com/leduckhai/MultiMed
This task aims to extract named entities and entity types while further predicting segmentation masks of visual objects.
This the dataset for Every Language Counts: Learn and Unlearn in Multilingual LLMs.
We have developed a systematic method for constructing large text annotated image databases designed for exploiting vision-language modeling for image quality assessment and present the Text Annotated Distortion, Appearance and Content (TADAC) database containing over 1.6 million images annotated with texts about their semantic contents, distortion characteristics and appearance properties. We used existing labels or automatic image captioning to annotate the semantic content, designed a list of suitable textual phrases for describing the distortion characteristics, and developed automatic algorithms for computing the appearance properties and annotated these properties with suitable textual descriptions. The TADAC database is the first of its kind that is annotated with all three types of quality relevant texts to enable the learning of high level knowledge about all possible factors affecting image quality. TADAC has enabled the development of the first BIQA model (SLIQUE) that joint
The Synthetic Signature Bankcheck Images (SSBI) Dataset is the first publicly available dataset of bank check images with annotations for detecting handwritten components, including names, amounts, dates, and signatures. It also supports both writer-independent and writer-dependent signature verification tasks by providing labels for genuine and forged signatures and IDs of the signature authors.
The Situational Awareness Database for Instruct-Tuning (SAD-Instruct) is a dataset for dynamic task guidance. It contains situationally aware instructions for performing everyday tasks or completing scenarios in 3D environments. The dataset provides step-by-step instructions for these scenarios grounded in the situation's context. This context is defined through a scenario-specific scene graph that captures the objects, attributes, and environmental relations. The dataset is designed to enable research in grounded language learning, instruction following, and situated dialogue.
The collection is a crawl of the public W3C (*.w3.org) sites in June 2004. NIST distributed the collection to TREC participants. This is a subset of 174,311 emails from the data used at TRECENT 2005.
An annotated subset of the W3C email corpus.