You need to be a member of diydrones to add comments!

Join diydrones

Comments

  • Sorry for the slow reply - yes I can put up the imagery that I have left. I deleted a lot as it was taking up so much space and was uninteresting (to me!). It mainly consists of mpg videos taken from a PAL analog video camera. Is there  a place I can upload them to? If not I can put on my google drive and give you a link.

  • Developer

    Hi James,

    What I was primarily envisaging was collaboration on data. There are not many sets of UAV imagery available for researchers to use in testing algorithms. Making your raw images available would help there, and may encourage more people to work on shark recognition.

    Cheers, Tridge

  • Hi guys. Tridge,I am a member of PerthUAV, who are OBC entrants. We are using your excellent recognition code. The OBC is keeping me very busy and I haven't had time to do any work on shark spotting lately.

    I'm a beginner when it comes to image processing and recognition, so I'm not sure I could offer much help to your PhD student, although I'd be interested in their results.

    No need to worry about the MTK GPS, they just needed a shot of a plane for the cameras and that was all I had on hand. It's running openpilot, which doesn't do autonomous flight anyway.
  • Developer

    Nice work James.

    I'm supervising a PhD student at ANU who is working on image recognition algorithms for finding objects from UAVs. Get in touch if you want to collaborate at all. I assume you have seen the CanberraUAV image recognition system that we developed for the OBC?

    Cheers, Tridge

    PS: Get rid of that MTK GPS - you will regret using it. Eventually it will lead to a crash.

  • James, I was wondering what you are doing wrt image processing. I have been toying with the idea of modifying a handwriting recognition algorithm to see if it will do the job for shark recognition. Any thoughts or experience in this regard?

  • James, I think the platform can be refined to work. For the cost of a days helicopter hire, I am sure that it would become a non-issue quite quickly.

    However, I feel that the key is in the image recognition software. If that can be automated as you alluded to, then I think the problem will be a long way towards solved.

    Michael, thanks for the video of your balloon trials. It illustrated an effect of the wind that hadn't crossed my mind before. However, I am sure that with more resources at your disposal you would have overcome that eventually. 

  • Forgot to mention, there was a guy at my talk who was proposing to put up large balloons along the coast for shark spotting. He also wanted to sell advertising on them! Not sure when this was going to happen, but sounds l uke a good idea.
  • Nice video of the balloon. Good to see people trying the idea. As has already been mentioned, the wind is a big issue here in WA. A stabilised camera would help, and with a stills camera you can also set a very high shutter speed to help further. There are a number of ways to do that, even some android cameras that open up a whole world of options.

    I'd be happy to meet up with you guys to talk about ideas. I have not taken my project any further at this point, but have many other UAV projects on the go in my spare time! I'd hate to get bored! Of course the downside is I now do much less flying and much more developing.
  • with respect to balloon use - this was only conducted as my flying skills were less than reliable at the time.

    My initial proof of concept flights were with a UAP1 quadcopter and flown by a skilled pilot.  A video of this also on my channel. The revised hardware will be again fitted to a copter and maybe also fixed wing.

  • Michael, thanks for the video and Simon, really impressed how far you went with what you did.

    Whats your next step ?

    It does look like wind with a balloon is going to be an issue. 

    I hadn't really thought it would but your video proves it.

    Regarding photos vs video...

    I had thought a gimbled GoPro with a polarized filter could work.

    The GoPro can be set to record video and photos at the same time but not sure how you could get that back to a base station to analyze. I dont see any challenges with photo and video itself but rather the vehicle used to support it.

    In an ideal world you would broadcast the video and or photos so anybody with their phone on the beach and maybe an app could see the water for themselves.

    I know from first hand experience that water conditions can make it hard to spot anything in the water.

    My son and i have been evacuated out of the water because of a shark twice from our local beach.

    The first time (feb 2013) we could not see it.

    The second time (only a month ago) we could see the shark in the water from beach easily.

    Food for thought eh ;)

This reply was deleted.