I Got (Business) Schooled!

Last year I and 9 other scientists were awarded a UK Newton Fund-DOST Leaders in Innovation Fellowship. The fellowship gave us and tech transfer officers  crash courses in research commercialization in London, Oxford and at the Asian Institute of Management. We were taught how to make business plans, pitch a product in under 3 minutes, negotiate, estimate market size, and project revenues.

Coming from Science all of the lessons were new to me and I enjoyed every bit of it for the novel and unique insights. I'd like to remember all the lessons I learned so I'm writing them down.

1. From Nieves Confesor of AIM - information is the currency of negotiation. In most cases, the more you openly share information the more you can negotiate better terms for your company.

2. From Richard Cruz of Ideaspace -  Scientists and engineers tend to think that their solution to a problem is the best for the end-user. But does it really address the customer pain? Stripped down to essentials our solution might not be answering their actual pain at all. Gather deep market insights by getting feedback from your end user.

3. From Maoi Arroyo of Hybridigm - the end-user is not necessarily the one who pays. Marketing should be directed at the one who holds the purse strings. Innovation = Invention x Revenues. If it doesn't earn its just an invention.

4. From Ricardo Lim of AIM - the former AIM dean introduced us to Design Thinking. Empathizing is the first step.

This is not the last in the list. I will add as I remember some more, since, hey, its been a year. I'll end with an advice that is somewhat antithetical to how we do science but which makes sense when we are developing solutions for an end user- Fail fast. Fail often. Iterate.

Me and my LIF Cohorts together with Tech Transfer officers holding up our business plans. This was at the Royal Academy of Engineering in London last March 2015.

With THE Nieves Confesor at the Asian Institute of Management. Photo by KM Magtubo.

Putting subtitles in video

Recently, our new VIP graduate student member Henry Lee, Jr. started a machine learning course in our lab. Some of our alumni like Francis and BA expressed interest in attending but couldn't come. So upon Francis' suggestion we took a video of Henry's lecture.

Pizza to go with the lectures. Photo by Elexis Mae Torres.

The video turned out great but the audio was terrible. I'm sure going to buy a directional microphone for the succeeding lectures but for the meantime I thought of annotating his talk by putting subtitles to the video. So onto Google to search for a subtitle editor.

Surprisingly, every forum I read recommended Subtitle Edit, a free, open source software to do just that.  So I tried it.

First, our movie was captured by an HD Panasonic video camera and the file format is .MTS. The lecture ran a little over an hour so the video file came out in two files, the first set was 4GB the second 1.4GB. I converted the MTS file into MP4 using Avidemux (oh, that's another topic. The proper setting can be found here.).

At first the file won't open in SE, its error message says the video codecs are not around. It suggested a link for downloading codecs. I just clicked and installed away and Voila! MP4's now open.

The SE interface is very easy to use. Five stars. And I've started adding subtitles.

Le Subtitle Edit interface.

But into the 50th second of the video I got lazy, so the complete annotations will have to wait for another day. Meanwhile, we will just have to wear earphones when we watch the video.

How to Install Image Processing Design Toolbox in Scilab without using Atoms

Besides SIVP - or Scilab Image and Video Processing Toolbox, I'm a great fan of Harald Galda's Image Processing Design. It has morphological operations and blob analysis functions not found in SIVP. These functions are great for automating analysis of several regions of interest in a scene at once.

However, as of posting, IPD does not show up in Scilab's ATOMS Module Manager. So I looked for ways to install IPD bypassing ATOMS. Anyone who has done this in Linux or MacOS let me know. Here's what works for Windows:

1. Download IPD.
2. Open Scilab and type SCIHOME. SCIHOME gives the path where the modules are stored.
3. In Windows Explorer copy paste the SCIHOME path to open the folder. You should see a folder named 'atom'.
4. Unzip the IPD module in the 'atom' folder. The unzipped folder should have the name IPD and not the zip filename.
5. Search for 'loader.sce' in the IPD folder, open and execute in Scilab.
6. Check your Scilab console it should show something like:

Start IPD - Image Processing Design
Load macros
Load dependencies
Load gateways
Load help
Load demos

And that's how you know you've succesfully loaded IPD toolbox in your Scilab.

Movie poster fun

One time our lab had a Christmas Party with Epic Movies as the theme. So each subgroup had to come dressed as characters from that movie. My group had the great fortune of being "cast" in The Lord of the Rings. I decided me and my students will go all out. So besides making props like swords and axes and beard, we even made... Tadah!.. a theatrical teaser poster with our faces superimposed on the characters.

We took pictures of our faces posed in the same way as the characters. Then with the help of GIMP and the use of ratio and proportion to get the scale of the faces right, we made this:

and this:

Making it was so much fun, my students were all game.

Then I thought, what if I share the fun with my  AP 186 class (Image Processing), besides there's a nice image scaling lesson there. And so for three years running, the first activity of my students in AP 186 is to recreate a movie poster with their faces in place of the actors'.

And so this year, I present to you the promotional posters of AP 186 2016!!!

The steps in making your own poster are:

1. Take a picture of yourself posed in the same way as the character. Be sure to get the lighting right too.

2. Crop your face using GIMP or any image processing software of your choice.

3. In the poster and and in your cropped image, measure the pixel distance between two fiducial points, e.g. distance between the eyes, distance from ear to ear, or forehead to jaw. 

4. Take the ratio between  the fiducial distance of the poster  and your image. Use this ratio as the scale factor to reduce or enlarge the size of your face image to fit exactly in the poster face.

5. You can also use the ratios of your skin color and the poster skin color to make the skin tone of your face match that of the poster.

6. You can then feather or smoothen the edges so that your cut out face blends with the rest of the poster image.

Practice makes perfect, soon you'll be posting memes in 9gag! Which poster got your thumbs up?

Freeware I Like

My students know that I never use pirated software and I encourage them to do the same. If i do not respect intellectual property how do I expect other people to respect mine especially with my line of work. UP is big on IP nowadays but how can we prosper if our young students are not even aware that they are committing plagiarism when they cut and paste and not even cite their source? I keep reminding my students to acknowledge their sources of info or assistance in their reports in the hope that it becomes a habit with them. (Incidentally, my dear AP 186 students, you have been copying my figures in your blog reports and you didn't even cite me as the source! See Checklist No. 3.)

Since 2000 I've been looking for free educational software which I can let my students install. I learned about freeware back in 1999 when I acquired a new laptop which came with a StarOffice installer (I still have the CD with me!). This was the precursor of OpenOffice which lately is now LibreOffice.

OpenOffice was ahead of Microsoft when it came to creating pdfs of documents. By searching the net I learned how to write a book using OpenOffice.  I like its equation editor most of all. One can also perform simple calculations in tables inserted in the document.

Next I was fortunate to find Scilab, a Matlab-lookalike. It is quite suited for teaching image processing and if you have been following my blog you'll get a lot of image processing tips from me my students.

Before Adobe Photoshop CS2 became free I was already using GIMP for image processing. For video parsing I swear by Avidemux (stick to version 2.5 if you want to parse videos into image frames). For audio analysis Audacity has several processing tools.

Sketchup and Google Earth always get installed in all my laptops.

For compression, I use 7zip for those files which Windows cannot uncompress. For virtual checking of the innards of my computer I use SIW.

I'll probably add to the list (and links!) as I go along.

Checklist for Writing Manuscripts II

It is August. Yes. It means I'm suffering from migraine from reading student papers again. And to ease the pain I shall write about it once more and add to my Pet Peeves in Student Manuscripts. I last ended in No. 6 so now Dear Students Who Will Read This Post as Required Reading  if you do not want to incur my wrath when you have me check your manuscript, read and follow:

7. Equations not part of sentence (Irritation Level : 3 stars out of 5)
Equations should be treated as part of the sentence and NOT AS A FIGURE.
This is wrong:
"The hypotenuse of a triangle is shown in Equation 1. Here x is the length of the base and y is the height of the triangle.
h = sqrt { x^2 + y^2}   [1]"

This is right:
"The hypotenuse h of a triangle is given by
h = sqrt {x^2 + y^2}  [1]
where x is the length of the triangle's base and y is the height."

Need some higher authority to convince you to drop that habit? Here, read the Penn State's style for students guide on writing equations in reports.

8. Citation not part of sentence (Irritation Level : 3 stars out of 5)
Isn't it obvious that  if you write

"Such patterns appeared in the works of Rogers. [6] In other works the dominant texture is the set described by Joyce. [7] We have yet to see works on the patterns we found in the new artefact."

the references [6] and [7] appear to belong to the next sentence because they have been excluded by the period at the end of the previous sentence?

If you read journal papers at all and not just cite them blindly you'll be observant of the proper citation practice.

GRRRR

Checklist for Writing Manuscripts

Around August and the end of an academic year in UP (that's around Feb and March) I suffer headaches and high blood pressure from reading my students' manuscript - be it SPP papers or thesis manuscripts. Yes, reviewing student papers is an occupational hazard.

Paano ka naman hindi mapupundi e kung yung mga simpleng rules ng pagsulat ng paper, hindi sinusunod! (GRRR!@$)352$#*%@$20)

So to minimize stress, because right now is one of those episodes, I'll make here a list of my top 10 6 pet peeves in manuscripts and I'll tell my students to check the list before they send in their papers. (And mind you, I'm VERY busy right now, but I have to get this off my chest by writing about it. That's how seriously stressed I am.)

1. Figures too small. 
A picture speaks a thousand words but if it is too small to see the details or embedded labels, well forget it, I'd rather you describe the story in three paragraphs. Look, an A4 page with a one-inch margin on both sides has a width for text or figures which is 6 inches plus. 6 INCHES! So why is your figure only 2 inches wide? Maximize figures so their width fill the whole page, My Gulay! That's what figures are for, to be seen!

2. Poor grammar
Nothing gets my goat faster than poor English. Please, if you have problems, have your text checked by labmates or groupmates who have a better command of the language.

3. Poor  or no citation
Listen, this can get you kicked out. Seriously. You can be accused of plagiarism. You either cite the source or don't include that text or figure. If you will use a figure from another paper the appropriate way is to write the author and ask permission. Don't ever forget to cite the source of the figure. Even if you have cited the source in the text, you must also cite the reference in the caption. Visit this site for a primer on plagiarism. There are more sites, just google.

4. No storyline,  poor construction
Writing a paper, even a scientific one, is a form of story telling. So before you write, think about your best friend. Think how you would tell a story to your best  friend. Now think about how you will tell the story of your thesis or research paper to your best friend.  You'd explain it in a way that he or she gets the premise right? Good stories have an arc. Go make one.

5. A long intro- 7/8 about the work of others, 1/8 about your work
A good introduction answers four questions:

1. What are you doing?
2. Why are you doing it?
3. What has been done before?
4. What is new with your approach?

Yes I can see that you did your review of literature but if you think I am impressed, well no,  unless you can put your own work in the context of the work of others.

6. Turning in a revised paper without a detailed list of changes made
When the manuscript has passed through the initial review and you've made the changes, be kind to your suffering reviewer and give a  detailed list of the changes you made. In this way he or she doesn't have to go through the agony of reading your whole paper again but can go straight to check the revised parts. This will speed up the review process too.  Detail means you enumerate the points raised by the reviewer, then you indicate the page number, paragraph, and sentence where you inserted the recommended change.

I'll probably add to this list when I get cranky again. I'm ok now. Nakalabas na ng sama ng loob.