With a dramatic presentation, and no KTs in the 7th sem, we successfully concluded the first half. Now would start the actual implementation of the whole project in MATLAB. There was a lot to be done. And we had to start from scratch. The first thing I and Maurya tried was to make a GUI. We struggled for over 5 hours and had a celebration of sorts when we managed to display a “picture” in the GUI. But then we realized we were working on a different, older version of MATLAB. So we got the latest version and started off with it. Almost ashamed at my abysmally low participation in the project so far, I took the responsibility for the coding part. Of course I enjoyed it too. So whenever I would take a break from my CAT studies I would start fiddling around with MATLAB. Am I a nerd or what!!! The progress was slow, but every little piece completed motivated me to get along with another part. Gradually, quite a professional looking GUI, complete with menus, several parameter selectors and useful helpers emerged. The rest of the gang had no idea what I had been upto. They must have been pleasantly surprised to see the result when I explained to them all that I had done. Since I have never had the virtue they call humility, which I think is not much of a virtue at all, I showed off to the other group of four, i.e. Sagar, Dipesh, Pratik, Dipti. I even had little sessions with Dipesh where I tried to share with him how to display pictures in the GUI; basically to get him started with MATLAB. After that I was pretty sure Dipesh would be able to manage to put the other stuff by himself; which I do believe he did manage.

So with the GUI done, there were plenty of other technical things to be done. Skin tone detection, and background subtraction and face extraction… the challenges were aplenty. And we were dependent on Sagar for that. Or atleast his web-cam. Though we managed to get some pictures from digital cameras of friends, they were barely sufficient. We were no less diligent than a Microsoft, or a Macromedia, in terms of testing our software. And we needed pictures to test that.

There was this brief period when our arsenal didn’t have a webcam. So we used the next best thing we had, the pictures of numerous bollywood actresses, which the amorous side of me had painstakingly collected from the internet. I could have never guessed that those pictures would ever be put to a scientific use. It’s sad that Aishwarya Rai and Amisha Patel aren’t even aware about their monumental contribution to the world of science. Drooling over every picture we perused to use for the testing, we clipped the faces, bearing oh so seductive expressions, and put some of them in the database and used the others to test the recognition algorithm. The results were discouraging. Had the case been that Aishwarya was getting recognized as Amisha, we would have tweaked some parameters to try and fine tune the code. But when Maurya gets recognized as Amisha Patel, there is something terribly wrong. Either Mr. Matthew Turk and Mr. Alex Pentland were completely off the bulls eye and had hit the truck standing ten meters away from the dartboard with their theory of “face recognition using eigen faces” or our implementation had gone absolutely off track and we were hopelessly lost. That too, only a few weeks away from the last day of delivering the project. I, Maurya and Megha did find the outcome of Maurya being recognized as a hot woman humorous at some level. That humour was lost on Ritesh. His mind had relinquished control of all the five senses. He was not moving, staring at the computer screen as if his glare would make the software correctly recognize the faces. He sat there rigid in the chair, not uttering a word. Though the three of us were worried about the software not working as it was supposed to, at that point, it was Ritesh we were more worried about. We tried justifying that background subtraction had not been done, and the lighting in the actress pictures was unreal, etc. etc. All of those were true in some sense. But Ritesh was not taking any inputs at all. We decided to pack for the day and let Ritesh get back in his groove, where he probably could suggest a way out of the quandary we were in.

Having realized that pictures of actresses were not the right test pictures, we knew a webcam was required. We turned to the guy who was closest to us who had a web-cam, viz. Sagar. We got his webcam for a brief period of time, and he readily helped us as best as he could. We took pictures of all of us, in varied expressions, drawing moustaches and beards on people’s faces to make sure that the face recognition algorithm worked independent of the facial hair and expressions. And the person who stretched the software to all its computational limits was Maurya. He has been endowed with a set of lips that can twist and turn and stretch and contract as much as he desires. One picture with a grin spreading across the countenance, and then a sad expression with a mouth that arched to resemble the dome of a Victorian era, a smiley face which looked much smilier than the yahoo emoticons, and the next instant the tongue sticking out as if mocking the world at large. It needs to be kept in mind that all these pictures had to be matched to the picture of Maurya in the database, which bore the no-nonsense, straight face, look, as if he had just emerged from a funeral, or to state a situation that his MBA status deserves, as if he had just emerged from a business meeting. Our software matched them all. Yoohoo!!! We were delighted.

Interim, we had several interactions with the professors at our college. Patil wanted a month on month progress report. And his doubts would leave us floundering at times. We would keep adding code to the same file. He objected to that. In our second round of reporting, he complained “You have already shown me this file. Show me some other file.” We tried to tell him that we had written extra stuff down below. He wasn’t convinced. So we showed him “some other” file, which was called by the original code. And Patil was happy like a two year old who had just been given a candy by his grandpa. We once even had to go to his home to save our project on his home computer. I had already been victimized once, when he had called me to his place to discuss my poor attendance. This time, thankfully, he didn’t give his entire life story. Only the small part where he described how they used to cook some equivalent of “chapattis” at his home.

Our encounters with our guide too had been fairly entertaining. Once when we went to him, he cast an accusing look at us, and said, “Your project has already been done before.” As Maurya has rightly claimed in the IEEE Format section, at times like these, opening your mouth can only make a situation worse. So we waited. And our guide continued his bellowing. “I saw this advertisement. In which a stranger comes to a house. And a robot dog sees the face and recognizes that he is a stranger!!!” (I think, it was some bike ad he was referring too) “Then there is this movie of Amitabh Bachchan and Nana Patekar, in which Amitabh is disguised as a Sardar with a beard and turban. But a computer still recognizes him!!!!!!!!” Oh my good GOD! In my mind I couldn’t believe that this guy who just said all that was supposed to be guiding me. Somebody, anybody, tell my guide that if movies and ads were a depiction of the things that have been invented, then you could travel “back to the future”, that too, thrice. And there was an island somewhere out there with dinosaurs on them And there was a special knife in the possession of Mithun da which could slice a fired bullet in exact two and the two halves could separately kill two people. Not that we were taking the credit for INVENTING the face recognition technique, but talking about discarding the project because a director with a technological twist to his creative mind had included something like that in his movie, which only a handful of people, including my guide, had seen, was too ridiculous even by Parshavnath standards. That googly thrown by our guide was skillfully handled and thankfully we didn’t have to “scratch” our project.

Having done the skin tone detection and background subtraction, and with the GUI fully complete, only little fine tuning jobs had to be done. These little things can always be nagging. But the fact that you are so close to the end keeps you motivated. Very soon, we were done with the coding. And the only thing left was a report. Now that would have been extremely painful without the abilities and expertise of the multi-faceted Maurya in Microsoft Word. The report was done too! Then came the drudgery of printing, binding, etc. Things like getting a certificate, were a pain, but with the management skills unknowingly imbibed in us at PCT, we negotiated all of the little operational issues. Soon we were laden with a dashing report, a black hard bound volume, with golden embossing. Even today, that report brings back all the memories and pictures vividly to mind.