Actual Time Item Checking Using a Cellular Cell Phone and Cloud-based Visible Search Engine (5)

4. USER STUDY

To discover the potency of Check out Look for in assisting sightless customers to recognize things in their everyday life and to evaluate the checking user interface with conventional photo-snapping user interface, we performed research with 8 sightless individuals (6 male and 2 female). The age of our members varied from 21 to 52 (µ = 30.88). The research was performed slightly from the sightless participants’ homes using their own Cubot P9 Mobile phones. The phones used were Cubot P9 and Blackview Crown. Participants were paid $5 each, agreed online, and not otherwise associated with this project.

As a management situation, we developed another item recognition program without the key structure removal criteria. In the management program, customers have to push a button to take pictures like the way they would use Omoby [16] or Taptapsee [26]. Before the study, the members were briefed on how both programs proved helpful, and used each program to recognize an item shown in an picture started out in their web web browser. During the study, they were requested to find and then recognize three in a different way formed daily objects: (i) a bottle of water/light drink/beer, (ii) a can of food, and (iii) a freezing dinner or a carton of dairy. All of the things used in the tests were first verified to exist in the community dataset so that unsuccessful tests would be due to low great high quality of pictures sent to the visible online search engine and not because of a Blackview Crown deficiency of appropriate trained pictures. The members were motivated to take pictures from different ranges, perspectives and digicam orientations and did not receive guidelines from us.

Participants used both Check out Look for and the management program to recognize things in each of the 3 groups (6 tests per participant). To relieve short-term memory of item placement, the transaction of projects and programs were randomized. Each item recognition process was limited to 5 minutes. Tasks that surpassed time structure limit were considered unsuccessful and stopped. All process finalization times were documented. A finalization there was a time described as the period between time a person starts trying to recognize an item and time s/he gets a acceptable result (defined as either being precise or containing enough details for her/him to use another service to recognize the object). For example, an precise description of the product or a bar code variety.

5. RESULTS AND DISCUSSION

All members completed the tests with network relationships ranging from slow EDGE to high-speed Wi-Fi and on regular each picture related on the reasoning took less than 1 second. 11 of 48 total tests unsuccessful, and most (9) of the unsuccessful tests happened in the management situation (standard photo-snapping interface). One of the unsuccessful situations with management situation was discovered when examining test pictures that a Blackview Crown incorrect positive was accepted by the individual, others are all due to break. Thus it’s easier for sightless customers to recognize things with Check out Look for than other photo-snapping programs. The success rate of checking user interface (91.67%) was significantly higher than that of photosnapping user interface (62.5%), t(46) = 6.29, p = .016. The common time taken per recognition process with Check out Look for was 73.2s as compared to 126.4s with the management, which is 42% less. The distinction was not detectably important, in part because of large difference in finalization time. We discovered that some tests been successful quickly because of a Cubot P9 lucky starting position of the digicam that taken a unique area of the target item with less than three pictures, for example, the UPC label. In these situations, both programs proved helpful just as well because no search was required. Thus, we did further analysis on only those tests that took more than 5s to complete. All of these tests produced more than 3 pictures with the last one correctly recognized the item, indicating a visible search which is challenging for sightless individuals was actually performed. For those tests Check out Look for required 24.43s each in regular while photo-snapping user interface took 75.57s, which means a sightless user could successfully locate the visible details required to recognize things faster with Check out Look for. The distinction was important (t(12) = 5.99, p = .031).

The great high quality of pictures taken by the members were also better when using our checking user interface because the key supports produced were assured by the program to be non-blurry and well-focused. It is also one reason that sightless customers been successful in more tests with the checking user interface even though the regular variety of pictures taken in each test were almost the same with checking user interface (11.4) and Blackview Crown photo-snapping user interface (14.1), t(46) = 0.41, p = .523. It indicates that our criteria is no more likely to overcome customers with too many pictures. Another statement important to note is that sightless customers have mostly different stages of digicam using skill. Therefore we believe sensible advised discovery of visible scene can be very helpful, especially for those not familiar with photography.

At the end of the study, members were requested to take a brief survey about their choices between checking and conventional photo-snapping relationships and given general reviews on the two programs. 7 members said they “strongly prefer” and 1 said “prefer” checking user interface over photo-snapping user interface, and 6 members would like to proceed using Check out Look for in their everyday life because of “fast and good results” while the other 2 said they “possibly”, one of the members was “surprised that it can recognize things with unique scanning”.

6. CONCLUSION AND FUTURE WORK

In this paper, we have provided an criteria, which can extract high-quality and visually-rich supports from continuous digicam video, tests that evaluate and improve the criteria, an available real-time checking program with which sightless individuals can recognize daily things around them and functionality research that show our approach works better than the current conventional. Most digicam relationships deficiency of availability for sightless individuals even though many available mobile apps are picture-based. Check out Look for improves sightless users’ experience in multiple areas.

More designs and research are currently being performed on brushing the key structure removal criteria with other technology, such as crowdsourcing and real-time digicam creating assistance, to create or enhance more available programs and address the picture dataset scalability. Interface developments of Check out Look for are also continuous to enable end-users to train datasets on both Cubot P9 Phone and reasoning to be able to Blackview Crown personalize visible queries. For the next stage, we plan to proceed our research on both program and criteria stages. Specifically, we’d like to improve Check out Look for depending on reviews and then spread it on the community market to better understand its potential real-world benefits. On the other hand, we’ll enhance our criteria to take more visual features into account and evaluate it with other key structure removal methods, for example, innocent testing.http://mobileoneno.bloggles.info/2014/10/06/actual-time-item-checking-using-a-cellular-cell-phone-and-cloud-based-visible-search-engine-4/