hello beautiful faces (:

This week was full of a lot of adjustments. Unfortunately,  a lot of our progress during these past few weeks went down the drain because the results were very limited by the image. Previously, when I was able to segment the image into separate files that outlined the ICs perfectly, it was limited to the test image and the moment I would change the image file, the results varied and the accuracy was below 50%.  Because of the constraints that computer vision offers, we had to speak with Professor Crandall on another approach to our problem.

Professor Crandall, on Tuesday during our lab meeting, recommended us to try Google’s Cloud Vision program and compare its results to Tesseract, another object recognition application. He told us that because the ICs have given us a lot of trouble and visual comparison has not work, he suggested us to try matching by just using the text on the ICs. It sounds simple, but the texts on the ICs are very small and often faded away. Because of that, Tesseract did a horrible job reading the text on the IC, while Google’s OCR did a great job reading the text (roughly 85% accurate). Nonetheless, Professor Crandall wants us to continue experimenting with both OCR applications so we can include the results in our poster.

During the week, Professor Crandall gave us the opportunity to visit the Cyberinfastructure building at Bloomington’s campus. The main building was gorgeously designed, in my opinion, better than Luddy. However, most of our time there was spent underground in a separate building in which IU’s supercomputers rest and all of their network infrastructure lays. It was funny to me that every room has a system shut off button that can be easily clicked and probably ruin thousands of researchers’ lives. I have never seen a supercomputer in person, did not even know much about the topic either. Thank you Professor Crandall for giving our lab the opportunity to visit and learn more about the computer science world.

With two weeks being left, Emmanuel and I are starting to brainstorm what to include in our poster. Since we did not really complete much of our project, we plan on showing what we have been working on and where we got stuck at. Although our project was not complete, we, nonetheless, learned a lot these past eight weeks and has helped educate me more on research in computer science and the field of computer visions.

Excited for these next two weeks; and good luck to all the other REUs!

peace and love,

G