The AOLAH databases are contributions from Aswan faculty of engineering to help researchers in the field of online handwriting recognition to build a powerful system to recognize Arabic handwritten script. AOLAH stands for Aswan On-Line Arabic Handwritten where “Aswan” is the small beautiful city located at the south of Egypt, “On-Line” means that the databases are collected the same time as they are written, “Arabic” cause these databases are just collected for Arabic characters, and “Handwritten” written by the natural human hand.

Instructions: 

* There are two databases; first database is for Arabic characters, it consists of 2,520 sample files written by 90 writers using simulation of a stylus pen and a touch screen. The second database is for Arabic characters’ strokes, it consists of 1,530 sample files for 17 strokes. The second database is extracted from the previous accepted database by extracting strokes from characters.
* Writers are volunteers from Aswan faculty of engineering with ages from 18 to 20 years old.
* Natural writings with unrestricted writing styles.
* Each volunteer writes the 28 characters of Arabic script using the GUI.
* It can be used for Arabic online characters recognition.
* The developed tools for collecting the data is code acts as a simulation of a stylus pen and a touch screen, pre-processing data samples of characters are also available for researchers.
* The database is available free of charge (for academic and research purposes) to the researchers.
* The databases available here are the training databases.

Categories:
359 Views

The images containing honey bees were extracted from the video recorded in the Botanic Garden of the University of Ljubljana, where a beehive with a colony of the Carnolian Grey, the native Slovene species, is placed. We set the camera above the beehive entrance and recorded the honey bees on the shelf in front of the entrance and the honey bees entering and exiting the hive. With such a setup, we ensured a non-invasive recording of the honey bees in their natural environment. The dataset contains 65 images of size 2688 x 1504 pixels.

Categories:
137 Views

The dataset consists of two classes: COVID-19 cases and Healthy cases 

Instructions: 

Unzip the dataset

Categories:
1288 Views

The images containing honey bees were extracted from the video recorded  in the Botanic Garden of the University of Ljubljana, where a beehive with a colony of the Carnolian Grey, the native Slovene species, is placed. We set the camera above the beehive entrance and recorded the honey bees on the shelf in front of the entrance and the honey bees entering and exiting the hive. With such a setup, we ensured a non-invasive recording of the honey bees in their natural environment. The dataset contains 65 images of size 2688 x 1504 pixels.

Categories:
84 Views

This dataset consists of 2579 image pairs (5158 images in total) of wood veneers before and after drying. The high-resolution .png images (generally over 4000x4000) have a white background. The data has been collected from a real plywood factory. Raute Corporation is acknowledged for making this dataset public. The manufacturing process is well visualized here: https://www.youtube.com/watch?v=tjkIYCEVXko.

Instructions: 

There are two folders: "Dry" and "Wet". The "Wet" folder contains wet veneer images and the "Dry" folder dry veneer images. The files are numbered so that e.g. Wet_10 is an image of the same veneer as Dry_10, but the veneer has been dried in between.

Categories:
337 Views

This dataset contains three benchmark datasets as part of the scholarly output of an ICDAR 2021 paper: 

Meng Ling, Jian Chen, Torsten Möller, Petra Isenberg, Tobias Isenberg, Michael Sedlmair, Robert S. Laramee, Han-Wei Shen, Jian Wu, and C. Lee Giles, Document Domain Randomization for Deep Learning Document Layout Extraction, 16th International Conference on Document Analysis and Recognition (ICDAR) 2021. September 5-10, Lausanne, Switzerland. 

This dataset contains nine class lables: abstract, algorithm, author, body text, caption, equation, figure, table, and title.

Instructions: 

Image files are in png formats and the metafiles are in plain text. 

Categories:
299 Views

Without publicly available dataset, specifically in handwritten document recognition (HDR), we cannot make a fair and/or reliable comparison between the methods. Considering HDR, Indic script’s document recognition is still in its early stage compared to others such as Roman and Arabic. In this paper, we present a page-level handwritten document image dataset (PHDIndic_11), of 11 official Indic scripts: Bangla, Devanagari, Roman, Urdu, Oriya, Gurumukhi, Gujarati, Tamil, Telugu, Malayalam and Kannada.

Instructions: 

See the attached pdf in documentation for more details about the dataset and benchmark results. Cite the following paper if you use the dataset for research purpose.

Obaidullah, S.M., Halder, C., Santosh, K.C. et al. PHDIndic_11: page-level handwritten document image dataset of 11 official Indic scripts for script identification. Multimed Tools Appl 77, 1643–1678 (2018). https://doi.org/10.1007/s11042-017-4373-y

Categories:
598 Views

An offline handwritten signature dataset from two most popular scripts in India namely Roman and Devanagari is proposed here. 

Instructions: 

Writer identification dataset availability on Indic scripts is a major issue to carry forward research in this domain. Devanagari and Roman are two most popular and widely used scripts of India. We have a total of 5433 signatures of 126 writers, out of which 3929 signatures from 80 writers in Roman script and 1504 signatures from 46 writers in Devanagari scripts. Script-wise per writer 49 signatures from Roman and 32 signatures from Devanagari are considered making an average of 43 signatures per writer on whole dataset. We have reported a benchmark results on this dataset for writer identification task using a lightweight CNN architecture. Our proposed method is compared with state-of-the-art handcrafted feature based method such as gray level co-occurrence matrix (GLCM), Zernike moments, histogram of oriented gradients (HOG), local binary pattern (LBP), weber local descriptor (WLD), gabor wavelet transform (GWT) and it outperforms. In addition, few well known CNN arechitechture is also compared with the proposed method and it shows comparable performance. 

User guidance: The images are available in .jpg format with 24 bit color. The dataset is freely available for research work. Cite the following paper while using the dataset

Sk Md Obaidullah, Mridul Ghosh, Himadri Mukherjee, Kaushik Roy and Umapada Pal “Automatic Signature-based Writer Identification in Mixed-script Scenarios”, in 16th International Conference on Document Analysis and Recognition (ICDAR 2021), Lussane, Switzerland, 2021

Categories:
531 Views

The proposed dataset, termed PC-Urban (Urban Point Cloud), is captured with an Ouster LiDAR sensor with 64 channels. The sensor is installed on an SUV that drives through the downtown of Perth, Western Australia (WA), Australia. The dataset comprises over 4.3 billion points captured for 66K sensor frames. The labelled data is organized as registered and raw point cloud frames, where the former has a different number of registered consecutive frames. We provide 25 class labels in the dataset covering 23 million points and 5K instances.

Categories:
1156 Views

The dataset consists of 751 videos, each containing the performance one of the handball actions out of 7 categories (passing, shooting, jump-shot, dribbling, running, crossing, defence). The videos were manually extracted from longer videos recorded in handball practice sessions. 

Instructions: 

The directory scenes/ contains the videos in mp4 format with actions of interest performed in context of other players present in the scene. The files are arranged in subdirectories according to the action class of the action of interest. The directory actions/ contains the videos of performances of actions by single players isolated from the videos in scenes directory. The files are arranged in subdirectories according to the performed action class. Files are named so that the beginning of the name matches the original video from which the action is extracted. The directory player_detections/ contains the object detections for each frame in the videos.

Categories:
376 Views

Pages