Abstract
Air writing is a practice of writing the linguistic characters in free space utilizing the six degrees of freedom of hand motion. Researchers have proposed various methods to approach air-writing based on one or more dedicated external hardware, increasing production cost, and reducing hardware redundancy. To solve this problem, we propose a system that uses a generic webcam to detect and recognize the virtually written characters by a user as per their will, removing the dependence on external hardware and increasing the hardware redundancy. This system performs detection using HSV color space for creating the mask of the tracker or the tracking object and morphological operations for refinement of the mask. This system gives the user the freedom to select a writing object of any color, shape, or material for tracking purposes. The trajectory of the contour of the object’s mask is tracked and rendered on a virtual window. The air-written character is recognized using the convolutional neural network (CNN). The CNN is trained and tested on four different datasets, which are English handwritten characters of 26 different classes (A–Z), MNIST dataset with 10 different classes (0–9), Devanagari handwritten character dataset consisting of 36 different classes (ka-gya), and Devanagari handwritten digits consisting of 10 classes (0–9). The CNN is also tested on a custom dataset of air-written digits (0–9) consisting of 3 samples of each class by 20 different individuals. The accuracy achieved by the proposed system for isolated characters on respective datasets is 99.75%, 99.73%, 99.13%, 99.97%, and 99.81%.
Similar content being viewed by others
References
Microsoft Corporation. Azure Depth Platform. Time-of-flight camera. https://devblo-gs.microsoft.com/azure-depth-platform/time-of-flight-camera-system-overview.
Leap Motion Inc. Leap Motion. https://www.ultraleap.com.
Microsoft Corporation. Azure Kinect Developer Kit. 2019. https://docs.microsoft.com/en-us/azure/kinect-dk.
Itaguchi Y, Yamada C, Fukuzawa K. Writing in the air: facilitative effects of finger writing in older adults. PLoS ONE. 2019;14(12): e0226832. https://doi.org/10.1371/journal.pone.0226832.
Rispens J, Berckelaer IA. Hyperlexia: definition and criterion. Writ Lang Disord. 1991. https://doi.org/10.1007/978-94-011-3732-4_8.
Alam MS, Kwon K-C, Alam MA, Abbass MY, Imtiaz SM, Kim N. Trajectory-based air-writing recognition using deep neural network and depth sensor. Sensors. 2020;20(2):376. https://doi.org/10.3390/s20020376.
Chen M, AlRegib G, Juang BH. Air-writing recognition–part I: modeling and recognition of characters, words, and connecting motions. IEEE Trans Hum Mach Syst. 2016;46(3):403–13. https://doi.org/10.1109/THMS.2015.2492598.
Chen M, Alregib G, Juang BH. Air-writing recognition–part II: detection and recognition of writing activity in continuous stream of motion data. IEEE Trans Hum Mach Syst. 2016;46(3):436–44. https://doi.org/10.1109/THMS.2015.2492599.
Mohammadi S, Maleki R. Air-writing recognition system for Persian numbers with a novel classifier. Vis Comput. 2020;36:1001–15. https://doi.org/10.1007/s00371-019-01717-3.
Dash A, Sahu A, Shringi R, et al. AirScript–creating documents in air. In: 14th international conference on document analysis and recognition. 2017. p. 908–13.
Yanay T, Shmueli E. Air-writing recognition using smart-bands. Pervasive Mobile Comput. 2020;66: 101183. https://doi.org/10.1016/j.pmcj.2020.101183.
Kumar P, Saini R, Behera SK, Dogra DP, Roy PP. Real-time recognition of sign language gestures and air-writing using leap motion. In: 2017 fifteenth IAPR international conference on machine vision applications (MVA). 2017. p. 157–60.https://doi.org/10.23919/MVA.2017.7986825.
Yin Y, Xie L, Gu T, Lu Y, Lu S. AirContour: building contour-based model for in-air writing gesture recognition. ACM Trans Sensor Netw. 2019;15(4):25. https://doi.org/10.1145/3343855.
Arsalan M, Santra A. Character recognition in air-writing based on network of radars for human-machine interface. IEEE Sens J. 2019;19(19):8855–64. https://doi.org/10.1109/JSEN.2019.2922395.
Islam R, Mahmud H, Hasan MK, Rubaiyeat H. Alphabet recognition in air writing using depth information. 2016.
Luo Y, Liu J, Shimamoto S. wearable air-writing recognition system employing dynamic time warping. In: 2021 IEEE 18th annual consumer communications & networking conference (CCNC). 2021. p. 1–6. https://doi.org/10.1109/CCNC49032.2021.9369458.
Arsalan M, Santra A, Issakov V. Radar trajectory-based air-writing recognition using temporal convolutional network. In: 2020 19th IEEE international conference on machine learning and applications (ICMLA). 2020. p. 1454–59. https://doi.org/10.1109/ICMLA51294.2020.00225.
Chen H, Ballal T, Muqaibel AH, Zhang X, Al-Naffouri TY. Air writing via receiver array-based ultrasonic source localization. IEEE Trans Instrum Meas. 2020;69(10):8088–101. https://doi.org/10.1109/TIM.2020.2991573.
Choi J-W, Ryu S-J, Kim J-H. Short-range radar based real-time hand gesture recognition using LSTM encoder. IEEE Access. 2019. https://doi.org/10.1109/ACCESS.2019.2903586.
Roy PP, Kumar P, Patidar S, et al. 3D word spotting using leap motion sensor. Multimed Tools Appl. 2021;80:11671–89. https://doi.org/10.1007/s11042-020-10229-5.
Tolentino RE, Roderos CJ, Ballesteros NJ, Papag MR, Tengco AV, Sunglao AP. Recognition of air-drawn alphanumeric characters by applying slope orientation sequence matching algorithm using leap motion controller. In: 2019 IEEE 11th international conference on humanoid, nanotechnology, information technology, communication and control, environment, and management (HNICEM). 2019. p. 1–5. https://doi.org/10.1109/HNICEM48295.2019.9072879.
Bastas G, Kritsis K, Katsouros V. Air-writing recognition using deep convolutional and recurrent neural network architectures. In: 17th international conference on frontiers in handwriting recognition (ICFHR). 2020. p. 7–12. https://doi.org/10.1109/ICFHR2020.2020.00013.
Saini R, Kumar P, Patidar S, Roy P, Liwicki M. Trilingual 3D script identification and recognition using leap motion sensor. In: 2019 international conference on document analysis and recognition workshops (ICDARW). 2019. p. 24–8. https://doi.org/10.1109/ICDARW.2019.40076.
Alam MS, Kwon K-C, Imtiaz SM, Hossain M, Shinde R, Kim J, Kim N. Air-writing recognition using a fusion CNN-LSTM neural network. 2021.
Uysal C, Filik T. RF-Wri: an efficient framework for RF-based device-free air-writing recognition. IEEE Sens J. 2021;21(16):17906–16. https://doi.org/10.1109/JSEN.2021.3082514.
Ahlawat S, Choudhary A, Nayyar A, Singh S, Yoon B. Improved handwritten digit recognition using convolutional neural networks (CNN). Sensors. 2020;20(12):3344.
OpenCV. Contours: getting started. https://docs.opencv.org/3.4/d4/d73/tutorial_py_contours_begin.html.
OpenCV. Image Filtering. https://docs.opencv.org/4.x/d4/d86/group__imgproc__filter.html.
OpenCV. Contours Hierarchy. https://docs.opencv.org/4.x/d9/d8b/tutorial_py_contours_hierarchy.html.
Alzubi J, Nayyar A, Kumar A. Machine learning from theory to algorithms: an overview. J Phys Conf Ser. 2018;1142(1): 012012.
Sachin P. A-Z handwritten Data: Sourced from NIST and NMIST large datasets for handwritten letters. 2016. https://www.kaggle.com/sachinpatel21/az-handwritten-alphabets-in-csv-format.
Deng L. The mnist database of handwritten digit images for machine learning research. IEEE Signal Process Mag. 2012;29(6):141–2.
Acharya S, Pant AK, Gyawali PK. Deep learning based large scale handwritten Devanagari character recognition. In: Proceedings of the 9th international conference on software, knowledge, information management and applications (SKIMA). 2015. p. 121–26.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
Prabhat Kumar declares that he has no conflict of interest. Abhishek Chaudhary declares that he has no conflict of interest. Abhishek Sharma declares that he has no conflict of interest.
Ethical approval
This article does not contain any studies with human participants or animals performed by any of the authors.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This article is part of the topical collection “Soft Computing for Real Time Engineering Applications” guest edited by Kanubhai K. Patel and Pritpal Singh.
Rights and permissions
Springer Nature or its licensor holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Kumar, P., Chaudhary, A. & Sharma, A. A CNN Based Air-Writing Recognition Framework for Multilinguistic Characters and Digits. SN COMPUT. SCI. 3, 453 (2022). https://doi.org/10.1007/s42979-022-01362-z
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s42979-022-01362-z