Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
streetview_facade_dataset [2019/12/18 16:30]
waag
streetview_facade_dataset [2019/12/18 16:31] (current)
waag
Line 7: Line 7:
 {{ :​mapbox.png?​nolink&​600 |}} {{ :​mapbox.png?​nolink&​600 |}}
  
-The downside for this particular way of downloading images is that it is quiet easy to download a complete neighbourhood or city, but if you want to limit yourself to specific streets (in our case with more activity in shopping and leisure), you would still have to repeat the process a lot of times. After the processing a certain amount of data, you would have to pay Google for their services. Next to that you would have to manually remove a lot of residue images and, in our case where we are interested in just the part of the building’s façade that includes the shop or bar, you would also have to manually select only this part of the building and remove the rest. 
  
-In the future we would like to automatize this process using computer vision. In that case the image recognition would be able to recognize the bar- or storefront and make an automatic selection, but for our first set of experiments it made more sense to use a mechanical turk[1] to do so. This person was tasked to make screenshots manually within the Google StreetView environment. 
  
-However since it was important for our research to also save the location of the images, our programmer Jorrit Schaap wrote a script that automatically saves the screenshot under a file name that includes the GPS code.+The downside for this particular way of downloading images is that it is quite easy to download a complete neighborhood or city, but if you want to limit yourself to specific streets (in our case with more activity in shopping and leisure), you would still have to repeat the process a lot of times.  
 + 
 +After processing a certain amount of data, you would have to pay Google for their services. Next to that, you would have to manually remove a lot of residue images and, in our case where we are interested in just the part of the building’s façade that includes the shop or bar, you would also have to manually select only this part of the building and remove the rest. 
 + 
 +In the future, we would like to automatize this process using computer vision. In that case, the image recognition would be able to recognize the bar- or storefront and make an automatic selection, but for our first set of experiments,​ it made more sense to use a mechanical turk[1] to do so. This person was tasked to make screenshots manually within the Google StreetView environment. 
 + 
 +Howeversince it was important for our research to also save the location of the images, our programmer Jorrit Schaap wrote a script that automatically saves the screenshot under a file name that includes the GPS code.
  
  
  • streetview_facade_dataset.txt
  • Last modified: 2019/12/18 16:31
  • by waag