VirtuLator

Project Title

VirtuLator

Project Area of Specialization Augmented and Virtual RealityProject Summary

i. Virtual Reality is playing a vital part here as to give the more ecstatic feel ii. It is not an ordinary calculator like the ones having normal buttons and a screen for display iii. It possess digit classifiers and hand recognition which will be following the phenomenon of manually written digits which are then to be recognized by the system and hence, getting the calculation to be done iv. High focus on efficiency and accuracy is taken into consideration v. As this will be the new thing, therefore our target is to get at least 60% to 65% accuracy vi. Palm Segmentation and Background Cancellation modules are also going to be a part of it.

Project Objectives

Project objectives are as under: i. Translate all those outputs of calculations into electronic words in a text document format. The advantage of this electronic storage is that this data only requires far less physical space than the storage of the physical copies. ii. To provide an ease and convenience for blind people so that they cannot feel any kind of lack iii. To make a full use out of virtual environment iv. To provide users with accurate results v. To make users interaction more powerful

Project Implementation Method

Phase-1 – Research about models to be used As per our project is concerned, we have gone through various models but the models which are appropriate to make this approach possible are listed below: i. Hand Recognition ii. Palm Segmentation iii. Background Cancellation iv. Digit Classifier

Phase-II – Collecting the data points This phase is eventually the critical one as we need to specifically target and ensure that on what kind of data are we going to work on. There can as much amount of data points as possible depending on the needs of project. In our case, we will be working on 0 to 9 numbers and some mathematical signs like +, -, x, / and =.

Phase-III – Training the model with respect to data points collected As per our research, we found out that MNIST named dataset is very efficient for these kind of models to be trained efficiently. This dataset has a kit of variations regarding the data points to be used in our project. Furthermore, this dataset has also an upper hand of model integration, so in future if we plan to integrate some other models to this system, therefore we will not be facing any sort of issue.

Phase-1V – Integrating the models There are total four models to be used in the system. We will design and train them individually then after unit testing, we will be going to integrate them. Firstly, it is going to be an easy task as all the models are to trained on MNIST (same dataset), and we can also turn these models into different APIs in order to integrate them. An alternative for this is to use a numerical method of integrating using SciPy (a python framework).

Phase-V- Execution with respect to Testing The last step after integration of models is to execute them and observe how everything goes. This is to be done on the backgrounds with some complexities so as to get a clear idea before implementing it fully.

Benefits of the Project

1. Translate all those outputs of calculations into electronic words in a text document format. The advantage of this electronic storage is that this data only requires far less physical space than the storage of the physical copies.
2. To provide an ease and convenience for blind people so that they cannot feel any kind of lack 
3. To make a full use out of virtual environment
4. To provide users with accurate results
5. To make users interaction more powerful

Technical Details of Final Deliverable

Our system is totally dependent on the camera being used to detect the movement of hands as users can use their front camera of the device as well as they can configure their web-cameras too. High-resolution camera should be a priority to have a greater detection for the motion and to have a perfect expected output  

Moreover, it is to be taken into consideration that the user must not move both their hands following the same motion as this module is initially based on single hand moment-detection. Speech-recognition should work well so as to facilitate the blind users which are of a major consideration here.

Final Deliverable of the Project HW/SW integrated systemCore Industry ITOther IndustriesCore Technology Augmented & Virtual RealityOther TechnologiesSustainable Development Goals Good Health and Well-Being for PeopleRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	15000
Webcam	Miscellaneous	1	7000	7000
Microphone	Miscellaneous	1	3000	3000
Speaker	Equipment	1	5000	5000