camallen
commented
9 years ago Hi @alianos-
Sounds like you will want two images / subject, i.e. give me a subject and it has before and after image URL's for the client to display.
- I would set it up as a workflow with each land use type as a different task. Your example would have 3 marking tasks / image.
- If you want to allow the users to select the areas they see, then you can use subject sets to logically group the subjects. The workflow can have multiple subject sets, your client interface would then ask for subjects from a worfklow / subjet_set via the subjects end point.
- A classification stores the results of each completed task in the annotations array. Here is some mock data that the front end is using. Let us know if the API docs need updating.
- I addressed this via 2. Depending on how you setup the workflow (see the optional cellect params), you can have random, priority, pairwise random/ priority subject selection strategies. Normally you'd use random to get a good distribution of classifications across your data set but it depends on your use case. See this code for more details.
Unfortunately not, otherwise i would have pointed you at at! It's on our list of things todo. @aliburchard you might want to use this as a template question re how to setup a project in future docs.
alianos-
So I need some help on how to set up the project if you have time at some point. The task at hand presents the user with 2 aerial photographs of the same location for different years. We ask users to markup the percentage of land use types for each image. (e.g. 2005: 80% forest, 20% pasture, 2013: 75% forest, 25% pasture). There are are some additional flags they can raise on each image.
Finally, is there a something like a guide or a tutorial that I might have missed?