Gender Detection & Classification - face recognition dataset

The dataset is created on the basis of Face Mask Detection dataset

Dataset Description:

The dataset comprises a collection of photos of people, organized into folders labeled "women" and "men." Each folder contains a significant number of images to facilitate training and testing of gender detection algorithms or models.

The dataset contains a variety of images capturing female and male individuals from diverse backgrounds, age groups, and ethnicities.

https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F12421376%2F1c4708f0b856f7889e3c0eea434fe8e2%2FFrame%2045%20(1).png?generation=1698764294000412&alt=media" alt="">

This labeled dataset can be utilized as training data for machine learning models, computer vision applications, and gender detection algorithms.

💴 For Commercial Usage: Full version of the dataset includes 376 000+ photos of people, leave a request on TrainingData to buy the dataset

Metadata for the full dataset:

• assignment_id - unique identifier of the media file
• worker_id - unique identifier of the person
• age - age of the person
• true_gender - gender of the person
• country - country of the person
• ethnicity - ethnicity of the person
• photo_1_extension, photo_2_extension, photo_3_extension, photo_4_extension - photo extensions in the dataset
• photo_1_resolution, photo_2_resolution, photo_3_extension, photo_4_resolution - photo resolution in the dataset

OTHER BIOMETRIC DATASETS:

• Anti Spoofing Real Dataset
• Antispoofing Replay Dataset
• Selfies, ID Images dataset (5591 sets of 15 files)
• Selfies and video dataset (4 052 sets)
• Dataset of bald people, 5000 images

💴 Buy the Dataset: This is just an example of the data. Leave a request on https://trainingdata.pro/datasets to learn about the price and buy the dataset

Content

The dataset is split into train and test folders, each folder includes: - folders women and men - folders with images of people with the corresponding gender, - .csv file - contains information about the images and people in the dataset

File with the extension .csv

• file: link to access the file,
• gender: gender of a person in the photo (woman/man),
• split: classification on train and test

TrainingData provides high-quality data annotation tailored to your needs

keywords: biometric system, biometric system attacks, biometric dataset, face recognition database, face recognition dataset, face detection dataset, facial analysis, gender detection, supervised learning dataset, gender classification dataset, gender recognition dataset

Data Collection - TagX can provides the dataset based on following scenarios to train a biasfree face analysis and detection dataset- Single and multiple faces images Monk skin-tones covered All Facial angles included

Metadata for Face Images- We can provide following metadata along with the collected images Age Range Distance from camera Emotion State Accessories present(Eyeglasses, hat etc.) pose with the values of pitch, roll, and yaw.

Annotation of Face Images- We can provides annotation for face detection applications like Bounding box annotation, Landmark annotation or polygon annotation. We have a dataset prepared with bounding box annotation around faces for 30000 images.

This data is used in the second experimental evaluation of face smile detection in the paper titled "Smile detection using Hybrid Face Representaion" - O.A.Arigbabu et al. 2015.

Download the main images from LFWcrop website: http://conradsanderson.id.au/lfwcrop/ to select the samples we used for smile and non-smile, as in the list.

Kindly cite:

Arigbabu, Olasimbo Ayodeji, et al. "Smile detection using hybrid face representation." Journal of Ambient Intelligence and Humanized Computing (2016): 1-12.

C. Sanderson, B.C. Lovell. Multi-Region Probabilistic Histograms for Robust and Scalable Identity Inference. ICB 2009, LNCS 5558, pp. 199-208, 2009

Huang GB, Mattar M, Berg T, Learned-Miller E (2007) Labeled faces in the wild: a database for studying face recognition in unconstrained environments. University of Massachusetts, Amherst, Technical Report

Classify video clips with natural scenes of actions performed by people visible in the videos.

See the UCF101 Dataset web page: https://www.crcv.ucf.edu/data/UCF101.php#Results_on_UCF101

This example datasets consists of the 5 most numerous video from the UCF101 dataset. For the top 10 version see: https://doi.org/10.5281/zenodo.7882861 .

Based on this code: https://keras.io/examples/vision/video_classification/ (needs to be updated, if has not yet been already; see the issue: https://github.com/keras-team/keras-io/issues/1342).

Testing if data can be downloaded from figshare with wget, see: https://github.com/mojaveazure/angsd-wrapper/issues/10

For generating the subset, see this notebook: https://colab.research.google.com/github/sayakpaul/Action-Recognition-in-TensorFlow/blob/main/Data_Preparation_UCF101.ipynb -- however, it also needs to be adjusted (if has not yet been already - then I will post a link to the notebook here or elsewhere, e.g., in the corrected notebook with Keras example).

I would like to thank Sayak Paul for contacting me about his example at Keras documentation being out of date.

Cite this dataset as:

Soomro, K., Zamir, A. R., & Shah, M. (2012). UCF101: A dataset of 101 human actions classes from videos in the wild. arXiv preprint arXiv:1212.0402. https://doi.org/10.48550/arXiv.1212.0402

To download the dataset via the command line, please use:

wget -q https://zenodo.org/record/7924745/files/ucf101_top5.tar.gz -O ucf101_top5.tar.gz tar xf ucf101_top5.tar.gz

Selfies, ID Images Face Dataset

5 591 sets, which includes 2 photos of a person from his documents and 13 selfies. 571 sets of Hispanics and 3512 sets of Caucasians.

Photo documents contains only a photo of a person. All personal information from the document is hidden

💴 For Commercial Usage: Full version of the dataset includes 95 000+ photos of people, leave a request on TrainingData to buy the dataset

Metadata for the full dataset:

• assignment_id - unique identifier of the media file
• worker_id - unique identifier of the person
• age - age of the person
• true_gender - gender of the person
• country - country of the person
• ethnicity - ethnicity of the person
• photo_1_extension, photo_2_extension, …, photo_15_extension - photo extensions in the dataset
• photo_1_resolution, photo_2_resolution, …, photo_15_resolution - photo resolution in the dataset

Content

The dataset includes 2 folders: - 18_sets_Caucasians - images of Caucasian people - 11_sets_Hispanics - images Hispanic people

In each folder there are folders for every person in dataset. Files are named "ID_1", "ID_2" for ID images and "Selfie_1",..."Selfie_13" for selfies.

https://sun9-53.userapi.com/impg/dOFVs6YsLexi-rM0LBud5rc6bVsCQPq5bIvrnA/S-3MRJPo-IE.jpg?size=2560x1054&quality=95&sign=16fc124e8f61d43a371cf4f0712f6a14&type=album" alt="">

💴 Buy the Dataset: This is just an example of the data. Leave a request on https://trainingdata.pro/datasets to learn about the price and buy the dataset

TrainingData provides high-quality data annotation tailored to your needs.

keywords: biometric system, biometric dataset, face recognition database, face recognition dataset, face detection dataset, facial analysis, object detection dataset, deep learning datasets, computer vision datset, human images dataset, human faces dataset, machine learning, image-to-image, re-identification, id photos, selfies and paired id, photos, id verification models, passport, id card image, digital photo-identification

1. Specifications Product : Biometric Data

Data size : 200,000 ID

Race distribution : black people, Caucasian people, brown(Mexican) people, Indian people and Asian people

Gender distribution : gender balance

Age distribution : young, midlife and senior

Collecting environment : including indoor and outdoor scenes

Data diversity : different face poses, races, ages, light conditions and scenes Device : cellphone

Data format : .jpg/png

Accuracy : the accuracy of labels of face pose, race, gender and age are more than 97%

1. About Nexdata Nexdata owns off-the-shelf PB-level Large Language Model(LLM) Data, 3 million hours of Audio Data and 800TB of Annotated Imagery Data. These ready-to-go Biometric Data support instant delivery, quickly improve the accuracy of AI models. For more details, please visit us at https://www.nexdata.ai/datasets/computervision?source=Datarade

1. Specifications Format : 16kHz, 16bit, uncompressed wav, mono channel

Recording environment : quiet indoor environment, without echo

Recording content (read speech) : economy, entertainment, news, oral language, numbers, letters

Speaker : native speaker, gender balance

Device : Android mobile phone, iPhone

Language : 100+ languages

Transcription content : text, time point of speech data, 5 noise symbols, 5 special identifiers

Accuracy rate : 95% (the accuracy rate of noise symbols and other identifiers is not included)

Application scenarios : speech recognition, voiceprint recognition

1. About Nexdata Nexdata owns off-the-shelf PB-level Large Language Model(LLM) Data, 3 million hours of Audio Data and 800TB of Annotated Imagery Data. These ready-to-go Machine Learning (ML) Data support instant delivery, quickly improve the accuracy of AI models. For more details, please visit us at https://www.nexdata.ai/datasets/speechrecog?source=Datarade

Emotion recognition Dataset

Dataset comprises 199,955 images featuring 28,565 individuals displaying a variety of facial expressions. It is designed for research in emotion recognition and facial expression analysis across diverse races, genders, and ages.

By utilizing this dataset, researchers and developers can enhance their understanding of facial recognition technology and improve the accuracy of emotion classification systems. - Get the data

Examples of data

https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F22059654%2F22472a4de7d505ff4962b7eaa14071bf%2F1.png?generation=1740432470830146&alt=media" alt="">

This dataset includes images that capture different emotions, such as happiness, sadness, surprise, anger, disgust, and fear, allowing researchers to develop and evaluate recognition algorithms and detection methods.

💵 Buy the Dataset: This is a limited preview of the data. To access the full dataset, please contact us at https://unidata.pro to discuss your requirements and pricing options.

Metadata for the dataset

https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F22059654%2F8cfad327bf19d7f6fad22ae2cc021a5b%2FFrame%201%20(2).png?generation=1740432926933026&alt=media" alt=""> Researchers can leverage this dataset to explore various learning methods and algorithms aimed at improving emotion detection and facial expression recognition.

🌐 UniData provides high-quality datasets, content moderation, data collection and annotation for your AI/ML projects

The problem of training a deep neural network with a small set of positive samples is known as few-shot learning (FSL). It is widely known that traditional deep learning (DL) algorithms usually show very good performance when trained with large datasets. However, in many applications, it is not possible to obtain such a high number of samples. In the image domain, typical FSL applications are those related to face recognition. In the audio domain, music fraud or speaker recognition can be clearly benefited from FSL methods. This paper deals with the application of FSL to the detection of specific and intentional acoustic events given by different types of sound alarms, such as door bells or fire alarms, using a limited number of samples. These sounds typically occur in domestic environments where many events corresponding to a wide variety of sound classes take place. Therefore, the detection of such alarms in a practical scenario can be considered an open-set recognition (OSR) problem. To address the lack of a dedicated public dataset for audio FSL, researchers usually make modifications on other available datasets. This paper is aimed at providing the audio recognition community with a carefully annotated dataset for FSL and OSR comprised of 1360 clips from 34 classes divided into pattern sounds and unwanted sounds. To facilitate and promote research in this area, results with two baseline systems (one trained from scratch and another based on transfer learning), are presented.

Object recognition predominately still relies on many high-quality training examples per object category. In contrast, learning new objects from only a few examples could enable many impactful applications from robotics to user personalization. Most few-shot learning research, however, has been driven by benchmark datasets that lack the high variation that these applications will face when deployed in the real-world. To close this gap, we present the ORBIT dataset, grounded in a real-world application of teachable object recognizers for people who are blind/low vision. We provide a full, unfiltered dataset of 4,733 videos of 588 objects recorded by 97 people who are blind/low-vision on their mobile phones, and a benchmark dataset of 3,822 videos of 486 objects collected by 77 collectors. The code for loading the dataset, computing all benchmark metrics, and running the baseline models is available at https://github.com/microsoft/ORBIT-DatasetThis version comprises several zip files:- train, validation, test: benchmark dataset, organised by collector, with raw videos split into static individual frames in jpg format at 30FPS- other: data not in the benchmark set, organised by collector, with raw videos split into static individual frames in jpg format at 30FPS (please note that the train, validation, test, and other files make up the unfiltered dataset)- *_224: as for the benchmark, but static individual frames are scaled down to 224 pixels.- *_unfiltered_videos: full unfiltered dataset, organised by collector, in mp4 format.

The dataset comprises 16.7k images and 2 annotation files, each in a distinct format. The first file, labeled "Label," contains annotations with the original scale, while the second file, named "yolo_format_labels," contains annotations in YOLO format. The dataset was obtained by employing the OIDv4 toolkit, specifically designed for scraping data from Google Open Images. Notably, this dataset exclusively focuses on face detection.

This dataset offers a highly suitable resource for training deep learning models specifically designed for face detection tasks. The images within the dataset exhibit exceptional quality and have been meticulously annotated with bounding boxes encompassing the facial regions. The annotations are provided in two formats: the original scale, denoting the pixel coordinates of the bounding boxes, and the YOLO format, representing the bounding box coordinates in normalized form.

The dataset was meticulously curated by scraping relevant images from Google Open Images through the use of the OIDv4 toolkit. Only images that are pertinent to face detection tasks have been included in this dataset. Consequently, it serves as an ideal choice for training deep learning models that specifically target face detection tasks.

Snapshot img

Image Recognition Market Size 2024-2028

The image recognition market size is forecast to increase by USD 111.45 billion at a CAGR of 25.49% between 2023 and 2028.

The market is experiencing significant growth due to the rising instances of identity threats and the increasing popularity of cloud-based image analysis solutions. With the increasing use of digital platforms, the need to secure personal data and prevent identity fraud is becoming increasingly important. Additionally, image recognition is essential in various applications such as medical imaging, robotics, and autonomous vehicles.
Deep learning, including neural networks, are used for feature extraction and pattern recognition, while pre-trained models and training data are crucial for model accuracy. Cloud-based image recognition solutions offer cost-effective and efficient ways to analyze large volumes of data, making them a preferred choice for businesses and organizations. However, the high cost of deployment remains a challenge for smaller businesses and organizations, limiting their adoption of image recognition technology. Overall, the market is expected to grow steadily In the coming years as the demand for advanced security measures and efficient data analysis solutions continues to increase.

What will be the Size of the Image Recognition Market During the Forecast Period?

Request Free Sample

The market encompasses various applications, including visual inspection, image classification, automated driving, and robotics. Object detection technologies, such as Faster R-CNN and YOLOv3, leverage deep learning and convolutional neural networks to identify and classify objects within images or videos. Machine learning, trained on vast amounts of data, can extract features from edge and corner elements, enhancing accuracy.
Furthermore, transfer learning and pre-trained models facilitate the adoption of image recognition technology in diverse industries, from banking apps and mobile check deposits to healthcare, where it can detect tumors and broken bones. Computer vision technology also powers facial recognition, enhancing security and convenience in human-technology interactions. Image recognition software processes visual content, including people, text, actions, and picture elements, transforming pixels into meaningful information on a 2-dimensional grid.

How is this Image Recognition Industry segmented and which is the largest segment?

The image recognition industry research report provides comprehensive data (region-wise segment analysis), with forecasts and estimates in 'USD billion' for the period 2024-2028, as well as historical data from 2018-2022 for the following segments.

End-user

  Media and entertainment
  Retail and e-commerce
  BFSI
  IT and telecom
  Others


Deployment

  Cloud-based
  On-premise


Geography

  North America

    US


  Europe

    Germany


  APAC

    China
    India
    Japan


  Middle East and Africa



  South America

By End-user Insights

The media and entertainment segment is estimated to witness significant growth during the forecast period.

The media and entertainment industry holds a significant market share in image recognition technology. Facial recognition analysis in video surveillance systems at cinemas is a major driver, enhancing audience engagement and improving visitor experience through personalized promotions and information. This technology's adoption is increasing globally, fueling the market's growth in this sector. Computer vision, a critical component of image recognition, processes digital images and videos, enabling applications like facial recognition, object detection, and text recognition.

The market's growth is further driven by advancements in machine learning models, transfer learning, and deep learning toolboxes. Applications include self-driving cars, autonomous mobile robots, and augmented reality. In industries like manufacturing, image recognition is used for defect detection, while in healthcare, it aids In the diagnosis of diseases from medical images like MRIs, X-rays, and CT scans. The market's growth is expected to continue during the forecast period, driven by advancements in technology and its increasing applications.

Get a glance at the Image Recognition Industry report of share of various segments Request Free Sample

The media and entertainment segment was valued at USD 9.10 billion in 2018 and showed a gradual increase during the forecast period.

Regional Analysis

North America is estimated to contribute 36% to the growth of the global market during the forecast period.

Technavio's analysts have elaborately explained the regional trends and drivers that shape the market during the forecast period.

For more insights on the market share of various regions, Request Free Sample

Image recognition technology, a key component of c

1. Specifications Format : 8kHz, 8bit, u-law/a-law pcm, mono channel;

Environment : quiet indoor environment, without echo;

Recording content : No preset linguistic data，dozens of topics are specified, and the speakers make dialogue under those topics while the recording is performed;

Demographics : Speakers are evenly distributed across all age groups, covering children, teenagers, middle-aged, elderly, etc.

Annotation : annotating for the transcription text, speaker identification, gender and noise symbols;

Device : Telephony recording system;

Language : 100+ Languages;

Application scenarios : speech recognition; voiceprint recognition;

Accuracy rate : the word accuracy rate is not less than 98%

1. About Nexdata Nexdata owns off-the-shelf PB-level Large Language Model(LLM) Data, 3 million hours of Audio Data and 800TB of computer vision data. These ready-to-go Machine Learning (ML) Data support instant delivery, quickly improve the accuracy of AI models. For more details, please visit us at https://www.nexdata.ai/datasets/speechrecog?source=Datarade

(Always use the latest version of the dataset. )

Human Activity Recognition (HAR) refers to the capacity of machines to perceive human actions. This dataset contains information on 18 different activities collected from 90 participants (75 male and 15 female) using smartphone sensors (Accelerometer and Gyroscope). It has 1945 raw activity samples collected directly from the participants, and 20750 subsamples extracted from them. The activities are:

Stand➞ Standing still (1 min) Sit➞ Sitting still (1 min) Talk-sit➞ Talking with hand movements while sitting (1 min) Talk-stand➞ Talking with hand movements while standing or walking(1 min) Stand-sit➞ Repeatedly standing up and sitting down (5 times) Lay➞ Laying still (1 min) Lay-stand➞ Repeatedly standing up and laying down (5 times) Pick➞ Picking up an object from the floor (10 times) Jump➞ Jumping repeatedly (10 times) Push-up➞ Performing full push-ups (5 times) Sit-up➞ Performing sit-ups (5 times) Walk➞ Walking 20 meters (≈12 s) Walk-backward➞ Walking backward for 20 meters (≈20 s) Walk-circle➞ Walking along a circular path (≈ 20 s) Run➞ Running 20 meters (≈7 s) Stair-up➞ Ascending on a set of stairs (≈1 min) Stair-down➞ Descending from a set of stairs (≈50 s) Table-tennis➞ Playing table tennis (1 min)

Contents of the attached .zip files are: 1.Raw_time_domian_data.zip➞ Originally collected 1945 time-domain samples in separate .csv files. The arrangement of information in each .csv file is: Column 1, 5➞ exact time (elapsed since the start) when the Accelerometer & Gyro output was recorded (in ms) Col. 2, 3, 4➞ Acceleration along X,Y,Z axes (in m/s^2) Col. 6, 7, 8➞ Rate of rotation around X,Y,Z axes (in rad/s)

2.Trimmed_interpolated_raw_data.zip➞ Unnecessary parts of the samples were trimmed (only from the beginning and the end). The samples were interpolated to keep a constant sampling rate of 100 Hz. The arrangement of information is the same as above.

3.Time_domain_subsamples.zip➞ 20750 subsamples extracted from the 1945 collected samples provided in a single .csv file. Each of them contains 3 seconds of non-overlapping data of the corresponding activity. Arrangement of information: Col. 1–300, 301–600, 601–900➞ Acc.meter X, Y, Z axes readings Col. 901–1200, 1201–1500, 1501–1800➞ Gyro X, Y, Z axes readings Col. 1801➞ Class ID (0 to 17, in the order mentioned above) Col. 1802➞ length of the each channel data in the subsample Col. 1803➞ serial no. of the subsample

Gravity acceleration was omitted from the Acc.meter data, and no filter was applied to remove noise. The dataset is free to download, modify, and use.

More information is provided in the data paper which is currently under review: N. Sikder, A.-A. Nahid, KU-HAR: An open dataset for heterogeneous human activity recognition, Pattern Recognit. Lett. (submitted).

A preprint will be available soon.

Backup: drive.google.com/drive/folders/1yrG8pwq3XMlyEGYMnM-8xnrd6js0oXA7

Snapshot img

Facial Recognition Market Size 2024-2028

The facial recognition market size is forecast to increase by USD 11.82 billion, at a CAGR of 22.2% between 2023 and 2028.

The market landscape is experiencing substantial growth, leading to a significant increase in demand for advanced identity verification. Organizations are prioritizing security measures, resulting in a rising need for precise and efficient identity verification processes. Key market trends include technological advancements and the emergence of facial analytics, which enhance accuracy and efficiency.
However, the high cost of deployment remains a significant challenge, potentially limiting access for smaller businesses and organizations. Overcoming this hurdle is essential for fostering broader adoption of digital identity and security and ensuring sustained growth in the market, particularly in the coming years.
The facial recognition market is expanding, driven by AI facial recognition and biometric authentication technologies. These advancements support security surveillance, contactless identity verification, and emotion detection technology. Cloud-based facial recognition systems leverage video analytics for enhanced public safety applications and access control solutions. However, privacy regulations play a significant role in shaping market growth, ensuring secure and compliant implementation of these systems in various sectors.

What will be the Size of the Facial Recognition Market During the Forecast Period?

To learn more about the facial recognition market report, Request Free Sample

Facial recognition technology is widely used across sectors like education for attendance, healthcare for patient monitoring, and retail for access control. Biometric POS Terminals integrate facial recognition to enhance payment security and efficiency. This technology also supports banking and law enforcement with secure authentication and surveillance.
Companies and technology corporations are pioneering advancements in facial recognition and biometric access control systems, employing technologies like image recognition and speech recognition. Facial characteristics, including jawline and facial contours, are analyzed to authenticate individuals. The application of facial recognition technology extends to smart hospitality services, enhancing the overall customer experience. This technology offers enhanced security and efficiency across multiple industries.

How is the Facial Recognition Market Segmented?

The facial recognition market trends and analysis report provides comprehensive data (region-wise segment analysis), with forecasts and estimates in 'USD billion ' for the period 2024-2028, as well as historical data from 2018 - 2022 for the following segments.

Application Outlook 

  Identification
  Verification


Technology Outlook 

  3D
  2D
  Facial analytics


End-user Outlook 

  Media and entertainment
  BFSI
  Automobile and transportation
  Others


Region Outlook 

  North America

    The U.S.
    Canada





  Europe

    The U.K.
    Germany
    France
    Rest of Europe





  APAC

    China
    India


  South America

    Chile
    Argentina
    Brazil





  Middle East & Africa

    Saudi Arabia
    South Africa
    Rest of the Middle East & Africa

By Application

The market share growth by the identification segment will be significant during the forecast period. Facial recognition technology has emerged as a significant solution for identification and verification in various sectors. NEC Corporation, Microsoft, AWS, and other tech giants are leading the market with advanced facial recognition systems. KYC systems and digital payments are integrating facial recognition for secure authentication. Smartphone applications and physical security systems also utilize this technology for access control and surveillance.

Get a glance at the market share of various regions. Download the PDF Sample

The identification segment was valued at USD 3.04 billion in 2018. Facial recognition systems use facial features, such as jawline and unique identifiers, to authenticate individuals. These systems are widely adopted in public safety and physical security for identification and verification purposes. The transportation sector, particularly airports, has seen a significant increase in the adoption of facial recognition technology for entry/exit systems.
Sectors requiring strict access control and video surveillance, such as banking and law enforcement, are increasingly relying on facial recognition technology for identification and verification. Authentication techniques using facial recognition are more secure and efficient compared to traditional methods. The global market for facial recognition technology is expected to grow significantly due to its wide adoption in various sectors.

Regional Analysis

For more insi

People can recognise the faces of friends and family across a huge range of conditions, including across changes in age. Changes over time are, however, a problem for unfamiliar face processing. For example, our passports can be up to ten years old, and yet a viewer checking our identity must nevertheless make the match. Some people are particularly good at unfamiliar face processing - people known as super-recognisers are employed in some police and security settings. In addition, trained practitioners, known as forensic examiners, have been found to have an advantage at face matching. However, we do not know whether these people are especially good at generalising photos across age ranges and at matching/recognising age separated images.

This project investigated the ability to recognise familiar and unfamiliar faces across age-separated images using a series of behavioural experiments and computational modelling. The data provided here examined the ability to generalise across age in untrained control participants, super-recognisers and forensic examiners.

Description:

This mmWave Datasets are used for activity verification. It contains two datasets. The first dataset (FA Dataset) contains 14 common daily activities. This second one (EA Dataset) contains 5 kinds of eating activities. The data are captured by the mmWave radar TI-AWR1642. The dataset can be used by fellow researchers to reproduce the original work or to further explore other machine-learning problems in the domain of mmWave signals.

Format: .png format

Section 1: Device Configuration

• A commodity mmWave radar TI AWR1642, which integrates a 2 × 4 antenna array. The detailed information of it can be found at https://www.ti.com/product/AWR1642#:~:text=The%20AWR1642%20is%20an%20ideal,of%2076%20to%2081%20GHz.
• A TI DCA1000EVM data capture card is used to collect data from the mmWave device and send data to a laptop. The detailed information can be found at https://www.ti.com/tool/DCA1000EVM?keyMatch=DCA1000EVM.
• mmWave radar work at the frequency in the range of 77~81GHz. The sampling rate is fixed at 100 frames per second and each frame has 17 chirps.

Section 2: Data Format

We provide our mmWave data in heatmaps for the two datasets. The data file is in the png format. The details are shown in the following:

FA Dataset

• 11 participants are included in the FA Dataset.
• 14 activities are included in the FA Dataset.
• FA_d_p_i_u_j.png:
  • d represents the data to collect the data.
  • p represents the environment to collect the data.
  • i represents activity type index
  • u represents user id
  • j represents sample index
• Example:
  • FA_20220101_lab_2_3_4 represents the 4th data sample of user 3 of activity 2 collected in the lab

EA Dataset

• 2 participants are included in the EA Dataset.
• 5 activities are included in the EA Dataset.
• EA_d_p_i_u_j.png:
  • d represents the data to collect the data.
  • p represents the environment to collect the data.
  • i represents the activity type index
  • u represents the user id
  • j represents the sample index

Section 3: Experimental Setup

FA Dataset

• We place the mmWave device on a table with a height of 60cm.
• The participants are asked to perform fitness activity in front of a mmWave device with a distance of 2m.
• The experiments are conducted in a lounge with furniture (7.0m×4.0m).
• We recruit 11 volunteers aged from 20 to 44 with various heights from 162cm to 185cm and weights from 50kg to 86kg.

EA Dataset

• We place the mmWave device on a table with a height of 60cm.
• The participants are asked to eat with different utensils (i.e., fork, fork&knife, spoon, chopsticks, bare hand) in front of a mmWave device with a distance of 1m.
• The data are collected at an office with a size of (5.0m×3.0m).

Section 4: Data Description

• We develop a spatial-temporal heatmap to integrates multiple activity features, including the range of movement, velocity, and time duration of each activity repetition.

• We first derive the Doppler-range map of the users’ activity by calculating Range-FFT and Doppler-FFT. Then, we generate the spatial-temporal heatmap by accumulating the velocity of every distance in every Doppler-range map together. Next, we normalize the derived velocity information and present the velocity-distance relationship in time dimension. In this way, we transfer the original instantaneous velocity-distance relationship to a more comprehensive spatial-temporal heatmap which describes the process of a whole activity.

• As shown in Figure below, in each spatial-temporal heatmap, the horizontal axis represents the time duration of an activity repetition while the vertical axis represents the range of movement. The velocity is represented by color.

• We create 2 folders to store two dataset respectively. In FA folder, there are 14 subfolders, each contains repetitions from the same fitness activity. In EA folder, there are 5 subfolders, each contains repetitions with different utensils.

14 common daily activities and their corresponding folders

Folder Name	Activity Type	Folder Name	Activity Type
FA1	Crunches	FA8	Squats
FA2	Elbow plank and reach	FA9	Burpees
FA3	Leg raise	FA10	Chest squeezes
FA4	Lunges	FA11	High knees
FA5	Mountain climber	FA12	Side leg raise
FA6	Punches	FA13	Side to side chops
FA7	Push ups	FA14	Turning kicks

5 eating activities and their corresponding folders

Folder Name	Activity Type
EA1	Eating with fork
EA2	Eating with spoon
EA3	Eating with chopsticks
EA4	Eating with bare hand
EA5	Eating with fork&knife

Section 5: Raw Data and Data Processing Algorithms

• We also provide the mmWave raw data (.mat format) stored in the same folder corresponding to the heatmap datasets. Each .mat file can store one set of activity repetitions (e.g., 4 repetations) from a same user.
  • For example: EA_d_p_i_u_j.mat:
    • d represents the data to collect the data.
    • p represents the environment to collect the data.
    • i represents the activity type index
    • u represents the user id
    • j represents the set index
• We plan to provide the data processing algorithms (heatmap_generation.py) to load the mmWave raw data and generate the corresponding heatmap data.

Section 6: Citations

If your paper is related to our works, please cite our papers as follows.

https://ieeexplore.ieee.org/document/9868878/

Xie, Yucheng, Ruizhe Jiang, Xiaonan Guo, Yan Wang, Jerry Cheng, and Yingying Chen. "mmFit: Low-Effort Personalized Fitness Monitoring Using Millimeter Wave." In 2022 International Conference on Computer Communications and Networks (ICCCN), pp. 1-10. IEEE, 2022.

Bibtex:

@inproceedings{xie2022mmfit,

title={mmFit: Low-Effort Personalized Fitness Monitoring Using Millimeter Wave},

author={Xie, Yucheng and Jiang, Ruizhe and Guo, Xiaonan and Wang, Yan and Cheng, Jerry and Chen, Yingying},

booktitle={2022 International Conference on Computer Communications and Networks (ICCCN)},

pages={1--10},

year={2022},

organization={IEEE}

}

https://www.sciencedirect.com/science/article/abs/pii/S2352648321000532

Xie, Yucheng, Ruizhe Jiang, Xiaonan Guo, Yan Wang, Jerry Cheng, and Yingying Chen. "mmEat: Millimeter wave-enabled environment-invariant eating behavior monitoring." Smart Health 23 (2022): 100236.

Bibtex:

@article{xie2022mmeat,

title={mmEat: Millimeter wave-enabled environment-invariant eating behavior monitoring},

author={Xie, Yucheng and Jiang, Ruizhe and Guo, Xiaonan and Wang, Yan and Cheng, Jerry and Chen, Yingying},

journal={Smart Health},

volume={23},

pages={100236},

year={2022},

publisher={Elsevier}

}

KU-HAR: An Open Dataset for Human Activity Recognition (v 1.0)

Human Activity Recognition (HAR) refers to the capacity of machines to perceive human actions. This dataset contains information on 18 different activities collected from 90 participants (75 male and 15 female) using smartphone sensors (Accelerometer and Gyroscope). It has 1945 raw activity samples collected directly from the participants, and 20750 subsamples extracted from them.

Activities/ Classes

1. Stand ➞ Standing still (1 min)
2. Sit ➞ Sitting still (1 min)
3. Talk-sit ➞ Talking with hand movements while sitting (1 min)
4. Talk-stand ➞ Talking with hand movements while standing or walking(1 min)
5. Stand-sit ➞ Repeatedly standing up and sitting down (5 times)
6. Lay ➞ Laying still (1 min)
7. Lay-stand ➞ Repeatedly standing up and laying down (5 times)
8. Pick ➞ Picking up an object from the floor (10 times)
9. Jump ➞ Jumping repeatedly (10 times)
10. Push-up ➞ Performing full push-ups (5 times)
11. Sit-up ➞ Performing sit-ups (5 times)
12. Walk ➞ Walking 20 meters (≈12 s)
13. Walk-backward ➞ Walking backward for 20 meters (≈20 s)
14. Walk-circle ➞ Walking along a circular path (≈ 20 s)
15. Run ➞ Running 20 meters (≈7 s)
16. Stair-up ➞ Ascending on a set of stairs (≈1 min)
17. Stair-down ➞ Descending from a set of stairs (≈50 s)
18. Table-tennis ➞ Playing table tennis (1 min)

Contents of the .zip files

1.Raw_ time_ domian_ data.zip ➞ Originally collected 1945 time-domain samples in separate .csv files. The arrangement of information in each .csv file is: Column 1, 5 ➞ exact time (elapsed since the start) when the Accelerometer (col. 1) & Gyroscope (col. 5) output were recorded (in ms) Col. 2, 3, 4 ➞ Acceleration along X, Y, Z axes (in m/s^2) Col. 6, 7, 8 ➞ Rate of rotation around X, Y, Z axes (in rad/s)

2.Trimmed_ interpolated_ raw_ data.zip ➞ Unnecessary parts of the samples were trimmed (only from the beginning and the end). The samples were interpolated to keep a constant sampling rate of 100 Hz. The arrangement of information is the same as above.

3.Time_ domain_ subsamples.zip ➞ 20750 subsamples extracted from the 1945 collected samples provided in a single .csv file. Each of them contains 3 seconds of non-overlapping data of the corresponding activity. Arrangement of information: Col. 1–300, 301–600, 601–900 ➞ Accelerometer X, Y, Z axes readings Col. 901–1200, 1201–1500, 1501–1800 ➞ Gyro X, Y, Z axes readings Col. 1801 ➞ Class ID (0 to 17, in the order mentioned above) Col. 1802 ➞ length of each channel data in the subsample Col. 1803 ➞ serial no. of the subsample

Gravity acceleration was omitted from the Accelerometer data, and no filter was applied to remove noise. The dataset is free to download, modify, and use provided that the source and the associated article are properly referenced.

Use the .csv file of the Time_ domain_ subsamples.zip for instant HAR classification tasks. See this notebook for details. Use the other files if you want to work with raw activity data.

Citation Request

More information is provided in the following data paper. Please cite it if you use this dataset in your research/work: [1] N. Sikder and A.-A. Nahid, “**KU-HAR: An open dataset for heterogeneous human activity recognition**,” Pattern Recognition Letters, vol. 146, pp. 46–54, Jun. 2021, doi: 10.1016/j.patrec.2021.02.024

[2] N. Sikder, M. A. R. Ahad, and A.-A. Nahid, “Human Action Recognition Based on a Sequential Deep Learning Model,” 2021 Joint 10th International Conference on Informatics, Electronics & Vision (ICIEV) and 2021 5th International Conference on Imaging, Vision & Pattern Recognition (icIVPR). IEEE, Aug. 16, 2021. doi: 10.1109/icievicivpr52578.2021.9564234.

Cite the dataset as: A.-A. Nahid, N. Sikder, and I. Rafi, “KU-HAR: An Open Dataset for Human Activity Recognition.” Mendeley, Feb. 16, 2021, doi: 10.17632/45F952Y38R.5

Supplementary files: https://drive.google.com/drive/folders/1yrG8pwq3XMlyEGYMnM-8xnrd6js0oXA7

Conclusion

The dataset is originally hosted on Mendeley Data

The image used in the banner is collected from here and attributed as: Fit, athletic man getting ready for a run by Jacob Lund from Noun Projects

1. Overview With extensive experience in speech recognition, Nexdata has resource pool covering more than 50 countries and regions. Our linguist team works closely with clients to assist them with dictionary and text corpus construction, speech quality inspection, linguistics consulting and etc.

2. Our Capacity -Global Resources: Global resources covering hundreds of languages worldwide

-Compliance: All the Machine Learning (ML) Data are collected with proper authorization -Quality: Multiple rounds of quality inspections ensures high quality data output

-Secure Implementation: NDA is signed to gurantee secure implementation and Machine Learning (ML) Data is destroyed upon delivery.

1. About Nexdata Nexdata is equipped with professional Machine Learning (ML) Data collection devices, tools and environments, as well as experienced project managers in data collection and quality control, so that we can meet the data collection requirements in various scenarios and types. Please visit us at https://www.nexdata.ai/service/speech-recognition?source=Datarade

Task

Named Entity Recognition(NER) is a task of categorizing the entities in a text into categories like names of persons, locations, organizations, etc.

Dataset

Each row in the CSV file is a complete sentence, list of POS tags for each word in the sentence, and list of NER tags for each word in the sentence

You can use Pandas Dataframe to read and manipulate this dataset.

Since each row in the CSV file contain lists, if we read the file with pandas.read_csv() and try to get tag lists by indexing the list will be a string. ```

data['tag'][0] "['O', 'O', 'O', 'O', 'O', 'O', 'B-geo', 'O', 'O', 'O', 'O', 'O', 'B-geo', 'O', 'O', 'O', 'O', 'O', 'B-gpe', 'O', 'O', 'O', 'O', 'O']" type(data['tag'][0]) string You can use the following to convert it back to list type: from ast import literal_eval literal_eval(data['tag'][0] ) ['O', 'O', 'O', 'O', 'O', 'O', 'B-geo', 'O', 'O', 'O', 'O', 'O', 'B-geo', 'O', 'O', 'O', 'O', 'O', 'B-gpe', 'O', 'O', 'O', 'O', 'O'] type(literal_eval(data['tag'][0] )) list ```

Acknowledgements

This dataset is taken from Annotated Corpus for Named Entity Recognition by Abhinav Walia dataset and then processed.

Annotated Corpus for Named Entity Recognition is annotated Corpus for Named Entity Recognition using GMB(Groningen Meaning Bank) corpus for entity classification with enhanced and popular features by Natural Language Processing applied to the data set.

Essential info about entities:

• geo = Geographical Entity
• org = Organization
• per = Person
• gpe = Geopolitical Entity
• tim = Time indicator
• art = Artifact
• eve = Event
• nat = Natural Phenomenon