Sketch recognition

About: Sketch recognition is a research topic. Over the lifetime, 1611 publications have been published within this topic receiving 40284 citations.

A hybrid system for handprinted word recognition

Abstract: The authors describe an alternative approach to handprinted word recognition using a hybrid of procedural and connectionist techniques. They utilize two connectionist components, which are to concurrently make recognition and segmentation hypotheses, and to perform refined recognition of segmented characters. Both networks are governed by a procedural controller which incorporates systematic domain knowledge and procedural algorithms to guide recognition. A recognition method is presented whereby an image is processed over time by a spatiotemporal connectionist network. The scheme offers several attractive features including shift-invariance and retention of local spatial relationships along the dimensions being temporalized, a reduction in the number of free parameters, and the ability to process arbitrarily long images. Recognition results on a set of real-world isolated zip code digits are comparable to the best reported to date with a 96.0% recognition rate. A pilot implementation of the complete system, and results on overlapping and touching pairs of zip code digits are reported. >

Lexicon-driven recognition of one-stroke character strings in visual gesture

Abstract: Visual gesture recognition enables natural human-machine interaction, and writing characters in gesture can convey rich information of intention. However, the recognition of character strings in gesture is challenging because multiple characters are in a single-stroke trajectory without pen lift information. We propose a lexicon-driven approach for gesture character string recognition. Using a lexicon of words to guide character segmentation and recognition, and meanwhile combining the geometric scores of characters and redundant segments with character classification score, we can achieve fairly high recognition accuracy on one-stroke character strings. For experiments, we collected 1,590 gesture strings in 100 word classes of television channel names, and achieved string-level recognition accuracy over 80% on the test set.

Democratizing 3D dynamic gestures recognition

Abstract: Developing vision-based 3D gestures recognition systems requires strong expertise and knowledge in computer vision and machine learning techniques. As human-computer interaction researchers do not generally have a thorough knowledge of these techniques, we developed Gesta. Gesta is a tool that enables non-experts in vision computing and artificial intelligence techniques to rapidly develop a 3D gestures recognition system prototype and to support the gesture design process. This tool works with up to two Microsoft Kinects, and integrates the depth cameras calibration algorithm and the hidden Markov models classifier. The users can manage these complex functions through a simple graphical user interface, even if they do not have any expertise in computer vision and machine learning domains. A usability test with 12 researchers with experience in human-computer interaction has been conducted in order to evaluate the overall usability of this tool. Results demonstrate that the testers appreciated the Gesta tool which scored 88.9 points out of 100 in the Brooke's system usability scale.

Interpretation of Box-and-Pointer Diagrams in Classroom Learning Partner

Abstract: This thesis focuses on the design and implementation of a tool to interpret the meaning of box-and-pointer diagrams drawn in digital ink. The tool was designed to work with the Classroom Learning Partner presentation system. The interpreter was designed to use state of the art sketch recognition tools to recognize shapes, and state of the art text recognition tools to recognize text. Thesis Supervisor: Kimberle Koile, Ph.D. Title: Research Scientist

Associative Memory Model for Distorted On-Line Devanagari Character Recognition

Abstract: Machine and human interaction is very essential in today's scenario. This interaction would make search engines, social media, artificial intelligence, cognitive computing more interactive and user friendly. Handwriting recognition is the systematic process of identifying the characters, numbers and symbols present in the handwritten document. In the current work, a recognition model for digitizing handwritten Devanagari characters proposed. Auto associative recognition technique for Devanagari characters and numerals proposed in the current work by using classifiers. To solve recognition problem a dynamic model based on Hopfield neural network deployed. The model performs operation in parallel making it faster and optimal in solving recognition problem.