Country: United Kingdom
Year Submitted: 2017
University: The University of Reading
List of Team Members (with year of graduation): Callum Bramley (2017)
Faculty Advisers: Rachel McCrindle
Main Contact Email Address: email@example.com
The RESPOND application, developed with LabVIEW and CVI, helps dementia patients regain their independence and confidence by allowing them to recognise family, friends and any household object.
LabVIEW: Vision Development Module, Vision acquisition software
A laptop Webcam
The number of new dementia sufferers is increasing rapidly in the UK with the diagnosis rate expected to hit 1 million by 2025. (fig 1.) The cost to treat one of these patients alone each year is currently £27000. Because of this, by 2025 the cost to treat just new Dementia sufferers will be around £27 billion a year. The result of this is that health services are likely to discharge patients as quickly as possible.
Figure 1. The number of new cases of dementia is rising year on year.
Perhaps the most upsetting part of dementia is that sufferers are unable to recognise their own family and friends, yet they need to rely on these people to help them in their everyday life. This can lead to sufferers losing confidence and their independence. Another symptom that links directly to this is the inability to recognise objects. This can often be as simple as just not being able to find the word for an item.
The challenge was to build an application that will be able to help stroke patients at home after being discharged from a hospital, to aid them with the recognition of objects, family, and friends. The end product would be used on a tablet or phone, with a simple user interface, and fast recognition to ensure that is as easy to use as possible.
There are two parts of the application, the object recognition side utilises the “reverse image search” capabilities of Google to access the largest database of named images in the world. This is achieved by saving the image to be searched and passing it over to a text based executable created in LabWindows CVI. The result is then presented in a large text box, in an easy to read font. After consulting with a focus group, the functionality to also ‘speak’ the result was also added for patients who have difficulty reading.
The main part of the application however, is to add and recognise close family members and friends. This is achieved by using the nodal points of the face to take measurements from, this is commonly referred to as 2D recognition. These points are then converted into a numeric code which is saved in a TDMS file, this ensures that the data is stored and read quickly, as well as being well organised. The name and relation of the person are also added to the end of the numerical code. This numerical code is then compared to the current image when searching, and the closest match is then displayed in a text box. Again, the ability to speak the result is added, after consulting with the focus group.
Figure 2. The nodal points of the face
Figure 3. How the concept of nodal points can be utilised in Vision Assistant
LabVIEW allowed for the rapid development and prototyping of the code which uses the buttons on the user interface to generate simple events. These events are then made readable by reducing them into their own Sub VIs. In comparison to other coding languages, the user interface was able to be designed quickly and then adapted after speaking to a focus group using LabVIEW’s ‘drag and drop’ for user interfaces. The use of tabs within the user interface were also ideal for segmenting an application with two sections, such as this one.
The Vision Assistant included within the vision development module was incredibly beneficial, as it allowed the facial recognition algorithm to be coded incredibly quickly by plotting the nodal points intuitively.
As part of the application relied on a text based language, using LabWindows CVI was ideal as it integrates seamlessly into LabVIEW. This allowed for all of the application’s variables and parameters to be passed into this section of code easily.
Figure 4. Screenshot of the object recognition application
Figure 5. Further object recognition testing, showing that the application can work with complex items and multiple colours.
Figure 6. Screenshot of the facial recognition application
After 6 months of work the project is currently at the alpha stage, being able to fully function on a laptop or computer with a webcam. The next stage of the project is to develop this into a mobile or tablet application that can be purchased on the Apple app or Google Play Store. As mentioned above the facial recognition function uses 2D recognition, to provide more accurate recognition 3D algorithms could be used, however this would need significantly more development time. A side project which could also be completed is to introduce the application as a fixed unit on the front door, so that people living alone are able to recognise visitors before opening the door.
The majority of rehabilitation applications for dementia sufferers aim to improve the memory of sufferers. However, with this application they will be able to carry out everyday tasks far more easily, and regain their independence and confidence.