JOIN
Get Time
forums   
Search | Watch Thread  |  My Post History  |  My Watches  |  User Settings
View: Flat (newest first)  | Threaded  | Tree
Previous Thread  |  Next Thread
Utility of 8-band TIFF imagery | Reply
Hello,

I noticed that the test directory against which scoring is performed only contains the *_rgb.jpg and *_msrgb.jpg images. So it's unclear to me as to why the raw, multi-spectral TIFF imagery would be useful for model training, given that 8-band TIFF images aren't possible inputs at inference time.

Thanks,
Sam
Re: Utility of 8-band TIFF imagery (response to post by Supersak80) | Reply
There is test data in TIFF format in fMoW-full_test_v1.0.0.torrent and also in the fmow-full AWS bucket. During final testing you will have the option to choose between the TIFF or JPG formats (fmow-full or fmow-rgb data sets).
Re: Utility of 8-band TIFF imagery (response to post by walrus71) | Reply
Are you allowed to use both TIFF and JPG for the final testing?
Re: Utility of 8-band TIFF imagery (response to post by yellowsubmarine) | Reply
Yes, although it complicates testing a bit. But if you have a good reason to go this way then you can do so.
RSS