I'm glad to see someone trying this. I very (very) casually explored this same idea recently by just browsing the search features of Instagram and Flickr. The results weren't very promising so I concluded it probably wasn't worth investing further effort in the API. Looking at the Huntington Beach results for this morning on the swellphone website, I don't see a lot that would interest surfers:
And I would expect that to be one of the more popular spots.
When I was in school at UCSD years ago, I found scraping NOAA buoy data and loosely correlating it with personal experience was a good enough predictor as to whether it was worth the walk down to Blacks. But this is what surfline.com does at scale.
Incidentally, anyone know what happened to the site surfshot.com? I believe their model was just to send a couple photographers down the San Diego coast each morning and post pictures. Site has been around for years and worked well enough, I thought, but it hasn't updated since July.
Very cool usage. What is your next step? As you said, most but not all pictures have pictures of the water. Have you thought about using CV to detect water in the background as the next possible iteration to remove pictures that do not show conditions.
I thought about it, however it will make things more complicated. Currently I get the pics unfiltered from Instagram. If I need further processing, I need to process each image and cache the results, then compare the feed per spot for the next request for the same spot etc. It's not that it's not doable though.
I hate to say it, but this "solution" doesn't work. It was a good idea worth testing, but it's just not useful.
I'm not a surfer, but if surfers are dying for this kind of info, wouldn't they be willing to pay a subscription fee for access to an up-to-date surf report?
These days with TaskRabbit.com, you can hire someone to go take pictures at multiple times throughout the day. Let's say you can get your costs down to $25/day, which is about $750/month per surf "spot".
You would need to get 38 surfers paying $19.95 per month to break even. How many surfers do you think would be interested in this per location?
Why take just a picture? Have the guy take a video panning the entire ocean for 15? 30? seconds and then allow surfers to view video. Capture some random images of the video for people who don't want to look at video.
I agree that video is a better way to communicate the surfing conditions. This is why I added Youtube videos as well to the SwellPhone feed. However, Youtube seems to have far less geo tagged UGC media than Instagram.
I started surfing about a year ago and I initially thought of the crowd sourced conditions app but I realized that people hate giving that info away and that friends form SMS based surf checking networks that serve the same purpose.
I really like this approach and could see it working at popular, touristy spots, clever idea.
Not sure I am a fan of the geo targeted default. Wish I could see my default to Hawaii North shore. Also would be nice if I could click on the images and link to the instragram post. Dig the concept and the site. Well executed.
In the mobile app version there's an "Add to Favorites" functionality and a "Favorites" tab. You can add Pipeline and other North Shore as favorites and easily follow them. That's missing from the website though.
Nice creative solution. I looked through the website and found a couple spots with no photos - it might be worth putting a note saying 'no photos found', I reloaded the page because it just looked like it hadn't loaded an element.
Very cool. I tried to do something similar with FB checkins but for Snow Reports. Same problem. Using instagram to source the reports is an smart detour.
