VideoSurfs Up Dude   Radical or Bogus Video Search? videosurf

VideoSurf - Just another video Search Engine?

VideoSurf states they have taken video search to a whole new level by allowing computers to see inside of the content of videos. The search engine then bases its results on visual identification according to the site and offers up a selection of six selections for you to jump directly to in each video instead of having to watch it from the beginning.

Using patent-pending computer vision algorithms that can actually see the video's content, VideoSurf serves up more relevant results for your queries and offers a new, visual way for you to interact with the video set returned. You can refine your results based on the people who actually appear in the videos and pinpoint the specific moments you're interested in watching or sharing with your friends

Honestly, I am not exactly sure what to make of their statements. No matter what I searched on - monkey banana, Kate Beckinsale, Wolverine, Kris Kross - the words always showed up in the title and/or the description. So in that regard it seems like it is all text based search to me. Looking through the results of my search showed that many of the selected scenes did not even have Wolverine in the thumbnails. The same could be said about the monkey banana search. Of course Kate Beckinsale showed up in almost every thumbnail indexed scene because most of the videos were clips from her work or appearances on television or at awards shows.

Time to Run Some Tests

The search for Kate Beckinsale on VideoSurf offered up 461 results. The exact same search on Google Video offered up 3,970. VideoSurf appears to only pull videos from around 17 sites including big names like YouTube, Google Video, DailyMotion, MetaCafe, Hulu and others which might be a reason for the limited results. Of course they might state that it's because of the 'mind-bendingly complex' math that is done on the search results is only showing you what you want to see, or at least what it thinks you want to see.

Likewise the search for 'Kris Kross' on Video Surf brought in 103 matches while the same search on DailyMotion brought 32, of which a handful were not the teen rap group from the 80's which I was searching for. Google = 1,450 results.

What is unique about VideoSurf?

VideoSurf offers a top row of images that are clickable allowing you to sort your results by the person in the image. A search for Alan brought up Alan Rickman, Alan Jackson, Alan Arkin, Alan Alda and Steve Carrell and Steve Coogan. I was not exactly certain how those last two ended up in the list. Digging a little deeper I found that it was because Alan Arkin was mentioned in the description of the Steve Carrell videos and Steve Coogan plays Alan Partridge on British television.

ALSO â–ş  Beyond YouTube: Why MCNs are Looking to Other Video Platforms for New Opportunities

VideoSurf also offers up between six and eight frames from each video to give you a visual summary of the content. Of course going back to the Kris Kross search one of the sets of six frames was essentially, a purple one, a blue one, a white one, a yellow one etc. many of which had no faces or content of interest. There is a 'show only faces' option for the video summary that will cut down the number of frames to only show you those with faces. This could be as few as one frame like Kris Kross - I Missed the Bus, or it could be as many as six like Kris Kross - Warm it Up. The odd thing is that the latter has no faces in any of the frames shown.

The Summary

Like VideoSurf I thought I would show you a quick summary of my thoughts on their service which is still in beta at present.

VideoSurfs Up Dude   Radical or Bogus Video Search? tree3 Frame 1:

Their video 'vision' seems more like tunnel vision, excluding what might be accurate results. By offering you about 10% of what a wide ranging video search engine might do in the way of results, you may not find exactly what you are looking for, although they are trying to help you find it more quickly.

VideoSurfs Up Dude   Radical or Bogus Video Search? tree2 Frame 2:

The vision is also blurry in that I don't see it being of much value at present since it can't manage to pull exactly what you might be looking for and extraneous results still show up. Even the show only faces feature can't seem to discern if there is a face present or not in each frame.

VideoSurfs Up Dude   Radical or Bogus Video Search? tree Frame 3:

VideoSurf seems Black and White in that it appears to be missing something that would make it extremely useful. The social networking facet of the site is nothing new as DailyMotion, YouTube and even Google Video offer up a range of these features and they also offer more results on searches at times.

  • Christophor Rick

    Hello Tom,

    Thanks for the comments. I should think that it would be easy enough to make a computer understand the human face. After all, generally speaking, they are all the same. Two eyes within a certain distance, a nose, a mouth. Mostly eyebrows etc... If someone is going to do facial analysis I should think they could do it with a broad template and be able to match it then. Additionally another way I could see it being done was to have representative images of specific people and match the frame in the video against the image allowing for a small margin of error to take into account angle and such. It's an interesting field and one the government is, I'm sure, extremely interested in so they can process millions of minutes of video footage from public cameras to find the evildoers.

    Suddenly I have this sneaking suspicion that we're like in 1984 and we just don't know it.

    Anyway, I'm curious to see how it progresses and if it becomes effective

  • TomW


    Good analysis. The biggest problem with machine vision approaches (in addition apparently to the fact that the machine vision part doesn't seem to work very well) is that the technology still can't understand what it is looking at. It may know two frames are the *same* but still doesn't know who or what is in them. Most approaches I've seen to this are then to have humans tag the frames, but then you are back to the same problem any editorial approach has with respect to search-it simply can't scale with the volume of content.