Monday

This week,  the two main API toolkits we used were Google’s Cloud Vision API and Amazon’s Rekognition API.  Google’s API was more accessible and relevant, since they had available for the general public, unlike Amazon’s WS API which needed some sort of subscription, which was beyond our expectation. Google’s vision API environment had sample Optical Character Recognition and Label Recognition and Identification features which were able to both identify text and identify figures and event scenarios in an image. This was very interesting to tweak around with, and the key takeaway was to have our algorithm perform similar tasks, plus image matching.

They also provided code samples which can be used;

We continued the development of our algorithm, encountered many stumbling blocks, had many trial and errors, but moreover, persisted to better our already-present results.

Tuesday

During our weekly meeting, we reported on our progress with our project; the code was now able to generate separate images of different IC’s in an image, crop them out without the application of image distortion filters, and save them in different files. Compared to the previous code which crops even the littlest dot in a picture, this code, however, was able to overlook such attributes, and produced a close-to-perfect result – eg. given an image of 6 IC’s on a sheet of paper, 6/10 generated images will be relevant to your work. Dr. Crandall was impressed by our result, and his idea was to have us build a small iOS app which will perform our project’s highlighted objective.

Wednesday

Happy 4th of July!!!

Thursday

As I assimilated back into the work schedule after an odd day off in the middle of the week, I found that these last two days of the week are going to be quite busy. Apart from learning how to convert Python to Objective-C and/or Swift, I also started planning how Gerald and I would design our poster. The complexity of our project will require us not only to design a poster which highlights the main purpose of our project but to also make sure the impact of the poster happens during and after the poster session. Nonetheless, coding still goes on, and we’re hopeful for a successful deliverable at the end of the program.

Friday

Gerald, who is the main engine behind all our success with our code, and I continued with our algorithm development, going through a series of trial and error and desk checking. We are set to just relax and destress this weekend, and pick up where we left off this week, trying to perfect our image matching results with the segments of code we have already developed.

Have a Blessed Weekend!