Papers with Code · CC BY-SA 4.0

데이터셋 탐색

데이터셋의 설명·홈페이지를 찾고, 관련 태스크의 평가 기록을 살펴보세요.

관련 논문 56편

Body-Expression-Audio-Text

BEAT
BEAT 설명 보기

BEAT has i) 76 hours, high-quality, multi-modal data captured from 30 speakers talking with eight different emotions and in four different languages, ii) 32 millions frame-level emotion and semantic relevance annotations. Our statistical analysis on BEAT demonstrates the correlation of conversational gestures with \textit{facial expressions}, \textit{emotions}, and \textit{semantics}, in addition to the known correlation with \textit{audio}, \textit{text}, and \textit{speaker identity}. Based on this observation, we propose a baseline model, \textbf{Ca}scaded \textbf{M}otion \textbf{N}etwork \textbf{(CaMN)}, which consists of above six modalities modeled in a cascaded architecture for gesture synthesis. To evaluate the semantic relevancy, we introduce a metric, Semantic Relevance Gesture Recall (\textbf{SRGR}). Qualitative and quantitative experiments demonstrate metrics' validness, ground truth data quality, and baseline's state-of-the-art performance. To the best of our knowledge, BEAT is the largest motion capture dataset for investigating human gestures, which may contribute to a number of different research fields, including controllable gesture synthesis, cross-modality analysis, and emotional gesture recognition. The data, code and model are available on \url{https://pantomatrix.github.io/BEAT/}.

관련 논문 23편

BEAT-SMPLX-FLAME

BEAT2
BEAT2 설명 보기

We propose EMAGE, a framework to generate full-body human gestures from audio and masked gestures, encompassing facial, local body, hands, and global movements. To achieve this, we first introduce BEAT2 (BEAT-SMPLX-FLAME), a new mesh-level holistic co-speech dataset. BEAT2 combines MoShed SMPLX body with FLAME head parameters and further refines the modeling of head, neck, and finger movements, offering a community-standardized, high-quality 3D motion captured dataset. EMAGE leverages masked body gesture priors during training to boost inference performance. It involves a Masked Audio Gesture Transformer, facilitating joint training on audio-to-gesture generation and masked gesture reconstruction to effectively encode audio and body gesture hints. Encoded body hints from masked gestures are then separately employed to generate facial and body movements. Moreover, EMAGE adaptively merges speech features from the audio's rhythm and content and utilizes four compositional VQ-VAEs to enhance the results' fidelity and diversity. Experiments demonstrate that EMAGE generates holistic gestures with state-of-the-art performance and is flexible in accepting predefined spatial-temporal gesture inputs, generating complete, audio-synchronized results.

관련 논문 10편

ECG Heartbeat Categorization Dataset

ECG Heartbeat Categorization Dataset 설명 보기

This dataset is composed of two collections of heartbeat signals derived from two famous PhysioNet datasets in heartbeat classification, the MIT-BIH Arrhythmia Dataset and the PTB Diagnostic ECG Database. The number of samples in both collections is large enough for training a deep neural network. This dataset has been used in exploring heartbeat classification using deep neural network architectures, and observing some of the capabilities of transfer learning on it. The signals correspond to electrocardiogram (ECG) shapes of heartbeats for the normal case and the cases affected by different arrhythmias and myocardial infarction. These signals are preprocessed and segmented, with each segment corresponding to a heartbeat.

관련 논문 4편

Beatles

Beatles 설명 보기

This dataset includes the beat and downbeat annotations for Beatles albums. The annotations are provided by M. E. P. Davies et. al [1]. M. E. P. Davies, N. Degara, and M. D. Plumbley, “Evaluation methods for musical audio beat tracking algorithms,” in Technical Report C4DM-TR-09-06, Centre for Digital Music, Queen Mary University of London, 2009.

관련 논문 1편

Wing beats

Audio de mosquitos Aedes Aegypti
Audio de mosquitos Aedes Aegypti 설명 보기

Dataset Description: The dataset comprises audio recordings of the wing beats of Aedes aegypti mosquitoes and others, conducted in a semi-controlled environment. It encompasses approximately 18,706 seconds of recording, with properly labeled samples. Dataset Characteristics: - Data Type: Audio recordings. - Total Duration: Approximately 18,706 seconds. - Labeling: Labeled samples. - Environment: Semi-controlled. - Data Type: Audio recordings. - Total Duration: Approximately 18,706 seconds. - Species: Aedes aegypti and others.

관련 논문 1편

MOSQUITO WINGBEAT RECORDINGS

WINGBEATS
WINGBEATS 설명 보기

Context The database contains wav recordings from the same optical sensor inserted in-turn into six insectary boxes containing only one mosquito species of both sexes (about 200-300 flying mosquitoes in each cage). As the mosquitoes fly randomly through the sensor their wingbeat partially occludes the light from the transmitter to the receiver. The light fluctuation recorded is modulated by the wingbeat of the insect. The resulting signal is pseudo-acoustic, meaning that it sounds exactly like a microphone recording but has been acquired using optical means (however, not vision based). Insect Biometrics, in the context of our work, is a measurable behavioral characteristic of flying insects. Biometric identifiers are related to the shape of the body (main body size, wing shape, wingbeat frequency, pattern movement of the wings). Biometric identification methods use biometric characteristics or traits to verify species/sex identities when insects access endpoint traps following a bait. Content • 279,566 wingbeat recordings correctly labeled • 6 mosquito species (Ae. aegypti, Ae. albopictus, An. arabiensis, An. gambiae, Cu. pipiens, Cu. quinquefasciatus) • 3 genera of mosquito species (Aedes, Anopheles, Culex) Acknowledgements The data have been recorded at the premises of Biogents, Regensburg, Germany (https://www.biogents.com/) and with the help of IRIDEON SA, Spain (http://irideon.eu/ ). The data have been recorded using the device published in: Potamitis I. and Rigakis I., "Large Aperture Optoelectronic Devices to Record and Time-Stamp Insects’ Wingbeats," in IEEE Sensors Journal, vol. 16, no. 15, pp. 6053-6061, Aug.1, 2016. doi: 10.1109/JSEN.2016.2574762 The REMOSIS project that supported the creation of the database has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No 691131. We gratefully acknowledge the support of NVIDIA Corporation with the donation of a TITAN-X GPU used for training the deep learning networks used to classify mosquitoes’ spectra. Inspiration The point of having such recordings is to eventually embed optoelectronic sensors in automatic traps that will report counts, species and sex identity of captured mosquitoes. All species of this dataset can be dangerous as they are potential vectors of pathogens that cause serious illnesses. A widespread network of traps for insects of economic importance such as fruit flies and of hygienic importance such as mosquitoes allows the automatic creation of spatiotemporal maps and cuts down significantly the manual cost of visiting the traps. The creation of historical data can lead to the prediction of outbreaks and risk assessment in general. We provide code to read the data and extract the power spectral density signature of each wingbeat. We also extract Mel-scaled, filter-bank features. How about wavelets and time-varying autoregressive models? The starter code using top-tier shallow classifiers achieves a mean accuracy of 81-84%. Deep-learning performs better. Can you classify genus, perform clustering, apply transfer learning to spectral data? Come aboard and help humanity against killer mosquitoes!