Go to content Go to navigation Go to search

ICCV - Day 8

Final day of workshops, final day of ICCV.

I attended the 1st Workshop on Benchmarking Facial Image Analysis Technologies, which had an array of talks on the latest benchmarks and datasets for facial image analysis. Jonathon Phillips from NIST gave an overview of NIST face benchmarking and protocols including unveiling some upcoming video datasets for the next few years. The guys from NICTA Australia had a review on face biometric benchmarking, and some other interesting new datasets were presented for gender classification, 3D twins and expression challenges, partially occluded 3D faces and a couple of “in-the-wild” datasets for facial expression and facial landmarks (points).

It was a surprising fact that this is just the first time ever that a workshop on face image/video benchmarks has been organised, and I think this should continue as the face processing community is quite large and matured now.

It has all come to an end. Not sure whether I felt relieved (8 solid days is mind-taxing) or doleful (as this is all ending and there wasn’t enough time to know more people). But, sure it has been one heck of a conference!

The one most important thing I learn is how possible it is to do “good” research the “right” way. Exposure is often a requisite for knowing what is to be considered good and useful research and obviously, to know how to get it done, and there is no better place to be exposed to that than to go to a top conference. Also, close collaborations within research groups (given a professor leads ably) and between different research groups is such a powerful thing that these people seemed to practice it all the time, which is exactly what the research environment in Malaysia terribly needs.

If you happened to be in an institution that promotes rapid gratification in the form of assessing so-called regimental performance indicators plucked out from thin air, I suggest that you take a step back from that (for your own good sake of putting the right foot into what real research is), for it’s worth every single bit of it. A paper in a top prestigious conference of any field is worth countless number of quick-fire mediocre journal papers.

This conference (as with other top conferences like CVPR, ECCV I believe) is the perfect launchpad for any budding computer vision researcher. You will come away highly motivated to do the next big thing in CV.

Meanwhile, I will be off to the Sagrada Familia tomorrow (second time going there, but this time, I must go in!) and maybe will try get myself lost in the alleys of Barcelona Old City. Adios!

Update: Here’s a quick skyline picture of the Sagrada Familia.

Leave your comments

  You can enclose *strong* for strong or _emphasis_ for emphasis
Find out more at Textile Help
This is not SPAM

Live comment preview


to be posted on 25 January 2021, 19:27

Subscribe to Tulips by Email

 » This article » 

Back to Archive list