Transfer Learning

Presents two high-quality large-scale CLS datasets based on existing monolingual summarization datasets. Source: NCLS: Neural Cross-Lingual Summarization

Presents half a million samples and structured meta-data to encourage further research and societal engagement. Source: OmniArt: Multi-task Deep Learning for Artistic Data Analysis

The Video-based Multimodal Summarization with Multimodal Output (VMSMO) corpus consists of 184,920 document-summary pairs, with 180,000 training pairs, 2,460 validation and test pairs. The task for this dataset is generating and appropriate textual summary of an article and choosing a proper cover frame from a video accompanying the article. Source: https://github.com/yingtaomj/VMSMO

FSDKaggle2019 is an audio dataset containing 29,266 audio files annotated with 80 labels of the AudioSet Ontology. FSDKaggle2019 has been used for the DCASE Challenge 2019 Task 2, which was run as a Kaggle competition titled Freesound Audio Tagging 2019. The dataset allows development and evaluation of machine listening methods in conditions of label noise, minimal supervision, and real-world acoustic mismatch. FSDKaggle2019 consists of two train sets and one test set. One train set and the test set consists of manually-labeled data from Freesound, while the other train set consists of noisily labeled web audio data from Flickr videos taken from the YFCC dataset. The curated train set consists of manually labeled data from FSD: 4970 total clips with a total duration of 10.5 hours. The noisy train set has 19,815 clips with a total duration of 80 hours. The test set has 4481 clips with a total duration of 12.9 hours. Source: https://labs.freesound.org/datasets/ Image Source: https://labs.freesound.org/datasets/

PGDP5K is a dataset consisting of 5000 diagram samples composed of 16 shapes, covering 5 positional relations, 22 symbol types and 6 text types, labeled with more fine-grained annotations at primitive level, including primitive classes, locations and relationships, where 1,813 non-duplicated images are selected from the Geometry3K dataset and other 3,187 images are collected from three popular textbooks across grades 6-12 on mathematics curriculum websites by taking screenshots from PDF books.

• The Sims4Action Dataset: a videogame-based dataset for Synthetic→Real domain adaptation for human activity recognition. * Goal : Exploring the concept of constructing training examples for Activities of Daily Living (ADL) recognition by playing life simulation video games. * ** Sims4Action dataset is created with the commercial game THE SIMS 4** by executing actions-of-interest within the game in a "top-down" manner. It features ten hours of video material of eight diverse characters and multiple environments. Ten actions are selected to have a direct correspondence to categories covered in the real-life dataset Toyota Smarthome [2] to enable the research of Synthetic→Real transfer in action recognition. * Two benchmarks :* Gaming→Gaming* (training and evaluation on Sims4Action) and Gaming→Real (training on Sims4Action, evaluation on the real Toyota Smarthome data [2]). * Main challenge: Gaming→Real domain adaptation While ADL recognition on gaming data is interesting from a theoretical perspective, the key challenge arises from transferring knowledge learned from simulated data to real-world applications. Sims4Action specifically provides a benchmark for this scenario since it describes a Gaming→Real challenge, which evaluates models on real videos derived from the existing Toyota Smarthome dataset . # References [1] Let's Play for Action: Recognizing Activities of Daily Living by Learning from Life Simulation Video Games. Alina Roitberg*, David Schneider*, Aulia Djamal, Constantin Seibold, Simon Reiß, Rainer Stiefelhagen, In International Conference on Intelligent Robots and Systems (IROS), 2021 (* denotes equal contribution.) [2] Toyota smarthome: Real-world activities of daily living. Srijan Das, Rui Dai, Michal Koperski, Luca Minciullo, Lorenzo Garattoni, Francois Bremond, Gianpiero Francesca, In International Conference on Computer Vision (ICCV), 2019.

SpaRTUN a dataset synthesized for transfer learning on spatial question answering (SQA) and spatial role labeling (SpRL). Recent research shows synthetic data as a source of supervision helps pretrained language models (PLM) transfer learning to new target tasks/domains. However, this idea is less explored for spatial language. We provide two new data resources on multiple spatial language processing tasks. The first dataset is synthesized for transfer learning on spatial question answering (SQA) and spatial role labeling (SpRL). Compared to previous SQA datasets, we include a larger variety of spatial relation types and spatial expressions. Our data generation process is easily extendable with new spatial expression lexicons. The second one is a real-world SQA dataset with human-generated questions built on an existing corpus with SPRL annotations. This dataset can be used to evaluate spatial language processing models in realistic situations. We show pretraining with automatically generated data significantly improves the SOTA results on several SQA and SPRL benchmarks, particularly when the training data in the target domain is small.

WWW Crowd provides 10,000 videos with over 8 million frames from 8,257 diverse scenes, therefore offering a comprehensive dataset for the area of crowd understanding.

As part of an ongoing worldwide effort to comprehend and monitor insect biodiversity, we present the BIOSCAN-5M Insect dataset to the machine learning community. BIOSCAN-5M is a comprehensive dataset containing multi-modal information for over 5 million insect specimens, and it significantly expands existing image-based biological datasets by including taxonomic labels, raw nucleotide barcode sequences, assigned barcode index numbers, geographical information, and specimen size. Every record has both image and DNA data. Each record of the BIOSCAN-5M dataset contains six primary attributes: RGB image DNA barcode sequence Barcode Index Number (BIN) Biological taxonomic classification Geographical information Specimen size

Unsupervised Domain Adaptation demonstrates great potential to mitigate domain shifts by transferring models from labeled source domains to unlabeled target domains. While Unsupervised Domain Adaptation has been applied to a wide variety of complex vision tasks, only few works focus on lane detection for autonomous driving. This can be attributed to the lack of publicly available datasets. To facilitate research in these directions, we propose CARLANE, a 3-way sim-to-real domain adaptation benchmark for 2D lane detection. CARLANE encompasses the single-target datasets MoLane and TuLane and the multi-target dataset MuLane. These datasets are built from three different domains, which cover diverse scenes and contain a total of 163K unique images, 118K of which are annotated. In addition we evaluate and report systematic baselines, including our own method, which builds upon Prototypical Cross-domain Self-supervised Learning. We find that false positive and false negative rates of the evaluated domain adaptation methods are high compared to those of fully supervised baselines. This affirms the need for benchmarks such as CARLANE to further strengthen research in Unsupervised Domain Adaptation for lane detection. CARLANE, all evaluated models and the corresponding implementations are publicly available at https://carlanebenchmark.github.io.

This dataset contains Bangla handwritten numerals, basic characters and compound characters. This dataset was collected from multiple geographical location within Bangladesh and includes sample collected from a variety of aged groups. This dataset can also be used for other classification problems i.e: gender, age, district. Source: BanglaLekha-Isolated: A Comprehensive Bangla Handwritten Character Dataset

This dataset is constructed and based on the online free-access fictions that are tagged with sci-fi, urban novel, love story, youth, etc. It is used for Writing Polishment with Smile (WPS) a task that aims to polish plain text with similes. All similes are extracted by rich regular expression, and the extraction precision is estimated as 92% by labelling 500 random extracted samples. It contains 5M samples for training and 2.5k for validation and test respectively. Source: https://github.com/mrzjy/writing-polishment-with-simile

This is a dataset for segmentation and classification of epistemic activities in diagnostic reasoning texts. Source: Analysis of Automatic Annotation Suggestions for Hard Discourse-Level Tasks in Expert Domains

By releasing this dataset, we aim at providing a new testbed for computer vision techniques using Deep Learning. The main peculiarity is the shift from the domain of "natural images" proper of common benchmark dataset to biological imaging. We anticipate that the advantages of doing so could be two-fold: i) fostering research in biomedical-related fields - for which popular pre-trained models perform typically poorly - and ii) promoting methodological research in deep learning by addressing peculiar requirements of these images. Possible applications include but are not limited to semantic segmentation, object detection and object counting. The data consist of 283 high-resolution pictures (1600x1200 pixels) of mice brain slices acquired through a fluorescence microscope. The final goal is to individuate and count neurons highlighted in the pictures by means of a marker, so to assess the result of a biological experiment. The corresponding ground-truth labels were generated through a hybrid approach involving semi-automatic and manual semantic segmentation. The result consists of black (0) and white (255) images having pixel-level annotations of where the stained neurons are located. For more information, please refer to Morelli, R. et al., 2021. Automating cell counting in fluorescent microscopy through deep learning with c-ResUnet. Scientific reports. https://doi.org/10.1038/s41598-021-01929-5. The collection of original images was supported by funding from the University of Bologna (RFO 2018) and the European Space Agency (Research agreement collaboration 4000123556).

Publicly available dataset in the hotel domain (50M versus 0.9M) and additionally, the largest recommendation dataset in a single domain and with textual reviews (50M versus 22M). Source: HotelRec: a Novel Very Large-Scale Hotel Recommendation Dataset

An annotated dataset is released to enable dynamic scene classification that includes 80 hours of diverse high quality driving video data clips collected in the San Francisco Bay area. The dataset includes temporal annotations for road places, road types, weather, and road surface conditions. Source: Dynamic Traffic Scene Classification with Space-Time Coherence

Multimodal Dyadic Behavior (MMDB) dataset is a unique collection of multimodal (video, audio, and physiological) recordings of the social and communicative behavior of toddlers. The MMDB contains 160 sessions of 3-5 minute semi-structured play interaction between a trained adult examiner and a child between the age of 15 and 30 months. The MMDB dataset supports a novel problem domain for activity recognition, which consists of the decoding of dyadic social interactions between adults and children in a developmental context. Source: MMDB

OSAI introduces OpenTTGames - an open dataset aimed at evaluation of different computer vision tasks in Table Tennis: ball detection, semantic segmentation of humans, table and scoreboard and fast in-game events spotting. It includes full-HD videos of table tennis games recorded at 120 fps with an industrial camera. Every video is equipped with an annotation containing the frame numbers and corresponding targets for this particular frame: manually labeled in-game events (ball bounces, net hits, or empty event targets) and/or ball coordinates and segmentation masks, which were labeled with deep learning-aided annotation models.

According to the WHO, World report on vision 2019, the number of visually impaired people worldwide is estimated to be 2.2 billion, of whom at least 1 billion have a vision impairment that could have been prevented or is yet to be addressed. The world faces considerable challenges in terms of eye care, including inequalities in the coverage and quality of prevention, treatment, and rehabilitation services. Early detection and diagnosis of ocular pathologies would enable forestall of visual impairment. One challenge that limits the adoption of a computer-aided diagnosis tool by the ophthalmologist is, the sight-threatening rare pathologies such as central retinal artery occlusion or anterior ischemic optic neuropathy and others are usually ignored. In the past two decades, many publicly available datasets of color fundus images have been collected with a primary focus on diabetic retinopathy, glaucoma, and age-related macular degeneration, and few other frequent pathologies. The challenge for which this dataset was introduced aimed to unite the medical image analysis community to develop methods for automatic ocular disease classification of frequent diseases along with the rare pathologies. The Retinal Fundus Multi-disease Image Dataset (RFMiD) consists of a total of 3200 fundus images captured using three different fundus cameras with 46 conditions annotated through adjudicated consensus of two senior retinal experts. To the best of the authors knowledge, the dataset, RFMiD represents the only publicly available dataset that constitutes such a wide variety of diseases that appear in routine clinical settings. This aforementioned challenge promoted the development of generalizable models for screening retina, unlike the previous efforts that focused on the detection of specific diseases.

RoboPianist is a benchmarking suite for high-dimensional control, targeted at testing high spatial and temporal precision, coordination, and planning, all with an underactuated system frequently making-and-breaking contacts. The proposed challenge is mastering the piano through bi-manual dexterity, using a pair of simulated anthropomorphic robot hands. The initial version covers a broad set of 150 variable-difficulty songs. Source: RoboPianist: A Benchmark for High-Dimensional Robot Control Image Source: RoboPianist: A Benchmark for High-Dimensional Robot Control

Huggingface Datasets is a great library, but it lacks standardization, and datasets require preprocessing work to be used interchangeably. tasksource automates this and facilitates reproducible multi-task learning scaling. Each dataset is standardized to either MultipleChoice, Classification, or TokenClassification dataset with identical fields. We do not support generation tasks as they are addressed by promptsource. All implemented preprocessings are in tasks.py or tasks.md. A preprocessing is a function that accepts a dataset and returns the standardized dataset. Preprocessing code is concise and human-readable.

The bipedal skills benchmark is a suite of reinforcement learning environments implemented for the MuJoCo physics simulator. It aims to provide a set of tasks that demand a variety of motor skills beyond locomotion, and is intended for evaluating skill discovery and hierarchical learning methods. The majority of tasks exhibit a sparse reward structure.

CQR is an extension to the Stanford Dialogue Corpus. It contains crowd-sourced rewrites to facilitate research in dialogue state tracking using natural language as the interface. Source: A dataset for resolving referring expressions in spoken dialogue via contextual query rewrites (CQR)

This is a dataset used to test deep learning-supported deep learning for fault diagnosis: - A digital twin model for a robot. - A synthetic data from the digital twin to train a deep learning-based fault diagnosis model. - A real dataset collected from the real robot to test the sim-to-real performance. Download the dataset from: https://nextcloud.centralesupelec.fr/s/7AR6aamBZNXcRM8/download

ExHVV is a novel dataset that offers natural language explanations of connotative roles for three types of entities -- heroes, villains, and victims, encompassing 4,680 entities present in 3K memes. Source: What do you MEME? Generating Explanations for Visual Semantic Role Labelling in Memes Image Source: https://arxiv.org/pdf/2212.00715v1.pdf

M2QA (Multi-domain Multilingual Question Answering) is an extractive question answering benchmark for evaluating joint language and domain transfer. M2QA includes 13,500 SQuAD 2.0-style question-answer instances in German, Turkish, and Chinese for the domains of product reviews, news, and creative writing. 40% of the data are unanswerable questions, 60% are answerable.

The dataset contains a total of 253,070 records, with 18 features. The features are categorized into four different types: Metadata, Primary Data, Engagement Stats, and Label. Under the Metadata category contains basic information about the channel and video, such as their unique identifiers, date and time of publication, and thumbnail URLs. The Primary Data category contains information about the title and description of the video. The "Processed" columns refer to the cleaned data after denoising, deduplication and debiased for further analysis. The Engagement Stats category contains data on user engagement metrics for each video. The Label category contains predefined auto labels, human annotated labels, and AI generated pseudo labels. Auto labels are labels that are automatically derived based on a review of their titles, descriptions, and thumbnails over time. Channels with consistently misleading, exaggerated, or sensationalized content were labeled as clickbait. Those focusing on factual information delivery without emotional appeals were labeled non-clickbait. Human labels are labels that are manually derived by volunteer human annotators and AI labels are labels that are generated by a fine-tuned AI model. The following table presents a detailed overview and definitions of the features. | Feature Type | Feature Name | Data Type | Definition | |----------------------------|----------------------|---------------|--------------------------------------------------------------| | Metadata | channel_id | string | ID of the YouTube channel | | Metadata | channel_name | string | Name of the YouTube channel | | Metadata | channel_url | string | URL of the YouTube channel | | Metadata | video_id | string | ID of the video | | Metadata | publishedAt | datetime | Date and time when the video was published | | Primary Data | title | string | Title of the video | | Primary Data (Processed) | title_debiased | string | Debiased title of the video | | Primary Data | description | string | Debiased description of the video | | Primary Data (Processed) | description_debiased | string | Description of the YouTube video without bias | | Metadata | url | string | URL of the video | | Engagement Stats | viewCount | int | Number of views the video has received | | Engagement Stats | commentCount | int | Number of comments on the video | | Engagement Stats | likeCount | int | Number of likes on the video | | Engagement Stats | dislikeCount | int | Number of dislikes on the video | | Metadata | thumbnails | string | URL of the thumbnail for the video | | Label | auto_labeled | string | Automatically labeled using manual review | | Label (Processed) | human_labeled | string | Labeled by human | | Label (Processed) | ai_labeled | string | Labeled by an AI model fine-tuned on human labeled data | ## Paper * Data in Brief: https://doi.org/10.1016/j.dib.2024.110239 * arXiv Link: https://arxiv.org/abs/2310.11465 ## Dataset * Mendeley: https://data.mendeley.com/datasets/3c6ztw5nft/ ## Citation ### MLA Al Imran, Abdullah, Md Sakib Hossain Shovon, and M. F. Mridha. "BaitBuster-Bangla: A Comprehensive Dataset for Clickbait Detection in Bangla with Multi-Feature and Multi-Modal Analysis." Data in Brief (2024): 110239. ### BibText @article{IMRAN2024110239, title = {BaitBuster-Bangla: A Comprehensive Dataset for Clickbait Detection in Bangla with Multi-Feature and Multi-Modal Analysis}, journal = {Data in Brief}, pages = {110239}, year = {2024}, issn = {2352-3409}, doi = {https://doi.org/10.1016/j.dib.2024.110239}, url = {https://www.sciencedirect.com/science/article/pii/S2352340924002105}, author = {Abdullah Al Imran and Md Sakib Hossain Shovon and M.F. Mridha}, keywords = {Bangla clickbait dataset, YouTube clickbait, Multi-modal clickbait dataset, Multi-feature clickbait dataset, Bangla natural language processing, User behavior modeling, Social Media Analysis}, abstract = {This study presents a large multi-modal Bangla YouTube clickbait dataset consisting of 253,070 data points collected through an automated process using the YouTube API and Python web automation frameworks. The dataset contains 18 diverse features categorized into metadata, primary content, engagement statistics, and labels for individual videos from 58 Bangla YouTube channels. A rigorous preprocessing step has been applied to denoise, deduplicate, and remove bias from the features, ensuring unbiased and reliable analysis. As the largest and most robust clickbait corpus in Bangla to date, this dataset provides significant value for natural language processing and data science researchers seeking to advance modeling of clickbait phenomena in low-resource languages. Its multi-modal nature allows for comprehensive analyses of clickbait across content, user interactions, and linguistic dimensions to develop more sophisticated detection methods with cross-linguistic applications.} }

The Cifar10Mnist dataset is created using CIFAR-10 and MNIST data sources. Since the CIFAR-10 training set consists of 50000 images and the MNIST training set contains 60000 digits, the first 50000 digits from MNIST are padded on top of the CIFAR-10 images after making them slightly translucent. A first training dataset is then obtained (50000 images). Furthermore, the remaining 10000 MNIST digits are padded on top of 10000 random CIFAR10 images (with a fixed seed). This gives the possibility of having a second training dataset of 60000 images. For the test set, the 10000 CIFAR-10 images are padded over the 10000 MNIST digits.

A large-scale dataset for recognizing kinship in multimedia which extend FIW with multimedia data (i.e., video, audio, and contextual transcripts). Source: Families In Wild Multimedia (FIW-MM): A Multi-Modal Database for Recognizing Kinship

There was no predefined dataset of party symbols to be usedas a benchmark. We curated a dataset from various nationaland regional websites owned by the ECI. The dataset consists of symbols (image files) of 49 National and State registered parties approved by the ECI. For each image of theoriginal party symbol, 18 different distortions and transformations were created as variations to the training data. Each image is of the dimension 180 x 180. The final labeled dataset consists of 931 images of party symbols with their corresponding party names as the labels.

A large-scale machine comprehension dataset (based on the COCO images and captions). Source: Understanding Image and Text Simultaneously: a Dual Vision-Language Machine Comprehension Task

A large scale, C2C marketplace e-commerce dataset.

The noun–noun compounds dataset created by Fares (2016) consists of compounds annotated with two different taxonomies of relations; that is, for each noun–noun compound there are two distinct relations, drawing on different linguistic schools. The dataset was derived from existing linguistic resources, such as NomBank (Meyers et al., 2004) and the Prague Czech-English Dependency Treebank 2.0 (PCEDT; Hajič et al., 2012). Source: Noun-Noun Compound Dataset

A large-scale dataset of user annotations on seven common photographic defects. Source: Learning to Detect Multiple Photographic Defects

The pic2kal benchmark for calorie prediction contains 308,000 images from over 70,000 recipes including photographs, ingredients and instructions, matched with nutritional information. Source: https://arxiv.org/abs/2011.01082 Image Source: https://github.com/phiresky/pic2kcal

The full version of ReefSet used in Williams et al. (2024). This dataset contains strongly labeled audio clips from coral reef habitats, taken across 16 unique datasets from 11 countries. This dataset can be used to test transfer learning performance of audio embedding models. This folder includes: - 57,084 WAV files that make up ReefSet_v1.0. Each file is 1.88 seconds in length, sampled at 16 kHz. All files have an associated label. - reefset_annotations.json, which contains the associated label, file ID, filename, data sharer, dataset and recorder type used for each sample in ReefSet_v1.0. This information is also indicated in the filename of each file. - 'reefset_labels_by_dataset.csv' provides a table containing the counts of each label class within each dataset.