I just updated it though so please try to login again.
I was able to successfully able to log in, so that problem should be resolved. Could you add hjcqcz@umsystem.edu and srk5cv@umsystem.edu?
Would there be certain trials or species of plants you would recommend for initial testing? I’m trying to narrow it down so I can process the data in bite-sized chunks. I am able to download the recipe files as jsons, but I don’t think the data is useful for the work that we’re trying to do. Do you know what the different recipes mean and how they relate to the trials?
I think I figured out how to switch among the different trials by hitting on the triple dashes button on the right.
I’m looking at some of the different trials, and I’m noticing that in some of the trials, the growth of multiple species is put in the same trial, some have empty containers, and others are obscured by the camera, say the cherry bella radish trial for Summer 2023, or the red romaine bok choy recipe for summer 2023. Additionally, some trials start with the plant already sprouting, like the dwarf tomato trial for summer 2023. It would take some time to select good candidate trials, and then parse the image data to select time-ordered images.
In terms of filtering out all of the data, the most basic thing I can think of is converting the images to grayscale, filtering the images, and detecting if there are any ‘white’ pixels from the filtered image to determine if the image has any plants in it. However, it probably wouldn’t fix some of the datasets having multiple plant species. I looked online to see if there was some sort of software that could automatically identify plant species, but I didn’t see any conclusive results.
Finally, is there a way I can download data off of the huroku app using some sort of api? Since there is a large number of photos, I’d like to see if there is a way to automatically download the files using some sort of python program to automate data processing.
I see that you posted this earlier on the marsfarm forums. I’m going to look at more in depth to see if it can answer my questions.
After doing some experimentation, I was unable to discover a way to login to the heroku web app using python to download the image data. Would there be some way to access the database directly so I could download the image data more efficiently?