Category: $-Family of Gesture Recognizers

Recently, we posted about a paper that Lisa co-authored with long-time collaborators, Radu-Daniel Vatavu and Jacob O. Wobbrock, that appeared at MobileHCI’2018. The paper presented some optimizations for our well-known $P gesture recognition algorithm to make it feasible to run on low-resource devices. The new algorithm is called $Q. For more on the paper, see our project page and online demo here, or the camera-ready version of the paper here.

In the meantime, Mobile’HCI was held in Barcelona, Spain, in September. During the conference, we discovered that our paper received an Honorable Mention for Best Paper award! Check it out on the ACM Digital Library here.

Congratulations to both of our co-authors for a great paper and a great acknowledgment by the community!

Read More

This summer, I have been working on a project related to the $-family of gesture recognizers. The $-family is a series of simple, fast, and accurate gesture recognizers designed to be accessible to novice programmers. $1 [1] was created by Wobbrock and colleagues, and INIT Lab director Lisa Anthony contributed to later algorithms, including $N [2] and $P [3]. My goal this summer was to implement my own versions of the $-family algorithms, and then to try them out on a new dataset that was collected from adults and children in a different context than previous datasets collected by the INIT lab.

The first step of my work on this project was to understand how the different algorithms of the $-family work. I examined the advantages and limitations of each recognizer in the $-family by reading the related research papers and playing around with existing implementations of the recognizers. After studying the recognizers, I created my own implementations of $1 and $P in Javascript by making a web application. I faced several challenges when implementing these algorithms. My first challenge was to decide in what form the gestures to be recognized should be taken as input (predefined point arrays or through a canvas where user-defined gestures can be given as input). Using this $1 implementation as a reference, I normalized each of the gestures, then computed the distance between the gestures and performed user-defined gesture recognition through a canvas. While implementing the algorithms, I followed a step by step approach so that I could evaluate whether each function was working before moving forward with recognition. In the process, I learned the importance of debugging the program to help pinpoint errors in my code more efficiently than trying to find the problems manually.

After completing the web applications, my next task was to recognize gestures from a dataset with XML files as input. I created another implementation of the $1 recognizer in Python to learn and explore another programming language. I was initially unsure how to read in the gesture data from XML files so I had to learn how to parse them. I used the pseudo code presented in the original $1 paper [1] as a guide to implement the algorithm. Resampling the points of the gesture before recognition was challenging. Every gesture needed to have the same number of resampled points for recognition. To solve the issues I encountered while preprocessing the gestures, I plotted the gestures using the matplotlib library from Python. Not only did visualising gestures help in that context, it also helped me to understand why some gestures were wrongly recognized, since they looked more like the other gestures than what they actually were. Solving these errors and getting a correct implementation gave me a great sense of achievement. After implementing the recognition algorithms, I learned how to run user-independent recognition experiments where I systematically varied the number of participants included in the training set. Then I ran those experiments to find out the accuracy of the algorithms that I implemented. Now, I am working on analyzing articulation features [4] [5] of a new set of gestures to help quantitatively investigate the difference between adult’s and children’s gestures in a new context.

I am a final year undergraduate computer science student from MIT, Pune, India working with the INIT lab as an REU student this summer, as part of the UF CISE IMHCI REU program. I have greatly enjoyed my time working in the INIT lab. One thing I have really enjoyed while I’ve been here is related to another project I worked on: interacting with an ocean temperature application on the PufferSphere, which is a large interactive spherical display. Through my experience in the INIT lab, I have been able to closely follow the different stages of the research process. I’ve added to my technical knowledge through improved understanding of gesture recognizers, and I’ve also learned the importance of being clear and concise in scientific writing. I am looking forward to continuing my work on this project and understanding new ways to improve children’s gesture interaction experiences.

References:

[1] Wobbrock, Jacob O., Andrew D. Wilson, and Yang Li. “Gestures without libraries, toolkits or training: a $1 recognizer for user interface prototypes.” Proceedings of the 20th annual ACM symposium on User interface software and technology. ACM, 2007.

[2] Anthony, Lisa, and Jacob O. Wobbrock. “A lightweight multistroke recognizer for user interface prototypes.” Proceedings of Graphics Interface 2010. Canadian Information Processing Society, 2010.

[3] Vatavu, Radu-Daniel, Lisa Anthony, and Jacob O. Wobbrock. “Gestures as point clouds: a $ P recognizer for user interface prototypes.” Proceedings of the 14th ACM international conference on Multimodal interaction. ACM, 2012.

[4] Anthony, Lisa, Radu-Daniel Vatavu, and Jacob O. Wobbrock. “Understanding the consistency of users’ pen and finger stroke gesture articulation.” Proceedings of Graphics Interface 2013. Canadian Information Processing Society, 2013.

[5] Shaw, Alex, and Lisa Anthony. “Analyzing the articulation features of children’s touchscreen gestures.” Proceedings of the 18th ACM International Conference on Multimodal Interaction. ACM, 2016.

Read More

We are excited to announced that there is a new member of the $-family of gesture recognizers! A paper on a new super-quick recognizer optimized for today’s low-resource devices (e.g., wearable, embedded, and mobile devices) that I (Lisa) co-wrote with my long-time collaborators, Radu-Daniel Vatavu and Jacob O. Wobbrock, will appear at the upcoming International Conference on Human-Computer Interaction with Mobile Devices and Services (MobileHCI’2018) conference. The paper extends the current best-performing, most robust member of the $-family, $P, using some clever code optimizations to short-cut much of the computation $P undertakes, and makes this recognizer, which we call $Q, blazingly fast and able to work in real-time on low-power devices. Here is the abstract:

We introduce $Q, a super-quick, articulation-invariant point-cloud stroke-gesture recognizer for mobile, wearable, and embedded devices with low computing resources. $Q ran up to 142× faster than its predecessor $P in our benchmark evaluations on several mobile CPUs, and executed in less than 3% of $P’s computations without any accuracy loss. In our most extreme evaluation demanding over 99% user-independent recognition accuracy, $P required 9.4s to run a single classification, while $Q completed in just 191ms (a 49× speed-up) on a Cortex-A7, one of the most widespread CPUs on the mobile market. $Q was even faster on a low-end 600-MHz processor, on which it executed in only 0.7% of $P’s computations (a 142× speed-up), reducing classification time from two minutes to less than one second. $Q is the next major step for the “$-family” of gesture recognizers: articulation-invariant, extremely fast, accurate, and implementable on top of $P with just 30 extra lines of code.

Radu will be presenting this work in the fall in Barcelona. Check out the camera-ready version of our paper here.

Read More

In our last post, we shared that we had a paper accepted to the ACM International Conference on Multimodal Interaction (ICMI) 2017, to be held in Glasgow, Scotland, UK. The paper was titled “Comparing Human and Machine Recognition of Children’s Touchscreen Gestures.” We just came back from the conference and are proud to announce that Alex Shaw, the INIT Lab PhD student who first-authored the paper, won Best Student Paper at the conference! Alex is co-advised by Dr. Lisa Anthony from the INIT lab and UF CISE professor Dr. Jaime Ruiz.

Congratulations, Alex!

Read More

In a previous post, we discussed our ongoing work on studying children’s gestures. To get a better idea of the target accuracy for continuing work in gesture recognition, we ran a study comparing human ability to recognize children’s gestures to machine recognition. Our paper, “Comparing Human and Machine Recognition of Children’s Touchscreen Gestures”, quantifies how well children’s gestures were recognized by human viewers and by an automated recognition algorithm. This paper includes our project team: me (Alex Shaw), Dr. Jaime Ruiz, and Dr. Lisa Anthony. The abstract of the paper is as follows:

Children’s touchscreen stroke gestures are poorly recognized by existing recognition algorithms, especially compared to adults’ gestures. It seems clear that improved recognition is necessary, but how much is realistic? Human recognition rates may be a good starting point, but no prior work exists establishing an empirical threshold for a target accuracy in recognizing children’s gestures based on human recognition. To this end, we present a crowdsourcing study in which naïve adult viewers recruited via Amazon Mechanical Turk were asked to classify gestures produced by 5- to 10-year-old children. We found a significant difference between human (90.60%) and machine (84.14%) recognition accuracy, over all ages. We also found significant differences between human and machine recognition of gestures of different types: humans perform much better than machines do on letters and numbers versus symbols and shapes. We provide an empirical measure of the accuracy that future machine recognition should aim for, as well as a guide for which categories of gestures have the most room for improvement in automated recognition. Our findings will inform future work on recognition of children’s gestures and improving applications for children.

The camera-ready version of the paper is available here. We will present the paper at the upcoming ACM International Conference on Multimodal Interaction in Glasgow, Scotland. We will post our presentation slides after the conference. See more information at our project website.

Read More

In previous posts, we have discussed our ongoing work on improving recognition of children’s touchscreen gestures. My paper, “Human-Centered Recognition of Children’s Touchscreen Gestures”, was accepted to ICMI 2017’s Doctoral Consortium! The paper focused on my future research plans as I continue to work on my doctorate. Here is the abstract:
Touchscreen gestures are an important method of interaction for both children and adults. Automated recognition algorithms are able to recognize adults’ gestures quite well, but recognition rates for children are much lower. My PhD thesis focuses on analyzing children’s touchscreen gestures, and using the information gained to develop new, child-centered recognition approaches that can recognize children’s gestures with higher accuracy than existing algorithms. This paper describes past and ongoing work toward this end and outlines the next steps in my PhD work.

We will post the camera-ready version of the paper soon. This year’s ICMI conference will be held in Glasgow, Scotland, in November.

I am beginning my 4th year as a PhD student at UF. I think that participating in the doctoral consortium at ICMI will be extremely helpful in continuing to develop a plan for my dissertation. I look forward to attending the conference.

Read More

We are currently continuing our work in gesture recognition by studying how well humans can recognize children’s gestures. We will compare human recognition rates to the rates of the automated recognition algorithms we used in our previous work. This will help us get an idea of how well humans are able to recognize children’s gestures. That way, we will have a good target accuracy for our future work on improving automated recognition of children’s gestures. Our future work will focus on improving the accuracy of recognizers for children’s gestures using the human recognition rate as the goal.

I am a rising 4th year Ph.D. student. Working on this project has helped me to better understand the ways that humans perceive gestures, which has led to some interesting discoveries on what kinds of gestures humans confuse. I look forward to applying this information to automated algorithms to improve recognition. Working on this project, I also learned how to use tools like Qualtrics and Amazon Mechanical Turk.

Read More

In a previous post, we discussed our ongoing work on studying children’s gestures. We studied a corpus of children’s and adults’ gestures and analyzed 22 different articulation features, which we are pleased to announce has been accepted for publication at the 2016 ACM International Conference on Multimodal Interaction (ICMI). Our paper, “Analyzing the Articulation Features of Children’s Touchscreen Gestures”, describes how children’s gestures differ from ages 5 to 10, and compares them to the features of adults’ gestures. This paper includes our project team: me (Alex Shaw) and Dr. Lisa Anthony. The abstract of the paper is as follows:

Children’s touchscreen interaction patterns are generally quite different from those of adults. In particular, it is known that children’s gestures are recognized by existing algorithms with much lower accuracy than those of adults. Previous work has qualitatively and quantitatively analyzed adults’ gestures to promote improved recognition, but this has not been done for children’s gestures in the same systematic manner. We present an analysis of gestures elicited from 24 children (age 5 to 10 years old) and 27 adults in which we calculate geometric, kinematic, and relative articulation features of the gestures. We examine the effect of user age on 22 different gesture metrics to better understand how children’s gesturing abilities and behaviors differ between various age groups. We discuss the implications of our findings and how they will contribute to creating new gesture recognition algorithms tailored specifically for children.

The camera-ready version of the paper is available here. We will present the paper at the upcoming ACM International Conference on Multimodal Interaction in Tokyo, Japan. We will post our presentation slides after the conference. See more information at our project website.

Read More

One of the projects our lab has been working on has been a qualitative analysis of the children’s gesture data from our MTAGIC project. We submitted an extended abstract about this work to CHI and it was accepted! In the extended abstract, we detail the tools we have applied thus far to study the children’s gestures, and summarize our findings. We also present an outline of our plans for continued work in this area. We will present our work as a poster at the conference. You can view the extended abstract that will be published here.

Read More

In my last post I discussed some work on the $-family of recognizers. Since then, I’ve been working on designing a standalone application for running gesture recognition experiments using the $-family. The application will also allow the user to design custom recognizers. I’ll also be using the $-family to run recognition experiments on the data we have collected in our MTAGIC project. By analyzing the results, we hope to be able to design better recognition algorithms for kids.

Working with the $-family has given me a great introduction to the field of gesture recognition. I have studied these algorithms as well as other work in the field, which will be important knowledge as I continue forward and begin writing my own gesture recognizers.

Read More