Asteroid Zoo Talk

Bad sets stuff.

  • MvGulik by MvGulik

    Mmm. Browsing around the general used "bad set/image/etc" tags it seems a bit of a jungle mix to me. One part that I think is not helping here is that there seems to be no related bad_stuff documentation/guidelines/etc. ... I do get the general "free to be used in any way the User's see fit" part (although I don't fully agree with it though.). In addition to that I also don't see any User way to clean that inevitable mixing up either.

    Ok. So much for my general personal gripe introduction. On to some data.

    What kind of 'distinct' bad stuff have I seen:

    • A) Sets with images of different sizes. (Not necessarily a bad/unusable thing. ... If it was not for the fact that the classify page itself centers the images in the used canvas, making them jump around. And not knowing how the classification clicks related to the images makes them kinda unclassifiable.)

    • B) Sets with garbled image data. (Those image usually also have a distinct brighter overall appearance.)

    • C) Sets with less than 4 images. (ignoring sets with more than 4 images.)

    • D) Sets with solid vertical streaks/bars in image. (Minor hiccup I think, as most, in general, seem still perfect classifiable.)

    Needless to say that a single set/image can have a mix of any of these issues.

    Bad set Tags encountered so far:

    1. bad
    2. bad_set
    3. badset
    4. bad_image
    5. bad-image
    6. badimage
    7. blank
    8. broken
    9. Incomplete
    10. misaligned
    11. missing_frames
    12. one_frame_set
    13. oneimage
    14. oneimageonly
    15. only1image
    16. retire_set
    17. unclassifiable

    I'm sure I missed some. But it illustrate the general issue.

    My point. ... well. I personally find this jungle a bit demotivating, and think it could use some cleanup, and some documentation/guidelines.


    (Just adding(edit) some additional personal notes/reminders.)

    • Hashtags are case insensitive.
    • "Hash-Tag" and "Hash_Tag" are both rendered as "Hash Tag" (for Search page, not in "Popular hashtags" list), but are processed as separate Hashtag cases.
    • Search: "Hash Tag" == "Hash-Tag".

    Posted

  • AstroTinker by AstroTinker in response to MvGulik's comment.

    MvGulik , I agree.
    Here is the list of hashtags I use or have used. I keep it in a separate file I can reference/update as needed. Obviously it has evolved as I have gained experience and researched problems on Asteroid Zoo. I only mark #bad-set for those I have checked the bad-set box, so #bad-noise, #bad-border, or #short-set often accompanies #bad-set, but might stand alone.

    This could be used as a basis for a glossary of hashtages, with links to examples.

    #asteroid


    #artifact not otherwise classified/unique

    #galaxy

    #globe-cluster


    #fmo fast moving object, streak on one image of set: satellite, meteor, aircraft?


    #ffd flat field defect - usually look like ghost stars, large or small black spots. most dark, some white, some may cause lens flare, some may have a semicircle or 'hat'. Will move in a regular pattern, often inverted T, inverted Y. If multiple, will move in formation,

    #ffd-white-spot white 'ghost' star

    #ffd-border fade/glare on a border, moving with any flat field defects, usually right border

    #ffd-haze haze with dark or sharp enough areas to see it moving in ffd pattern


    #cosmic-ray single hot pixel in one image of set

    #hot-pixel hot pixel in multiple images of set, should moves in ffd pattern

    #hot-pixel-column hot (white)pixel column in multiple images of set, moves in ffd pattern

    #bad-pixel-column pixel fail (black) in column, may be irregular/faded, moves in ffd pattern

    #star-bleed hard pixel bleed or 'rays' from star, verticle and/or horizontal
    may have black 'spot' near center from overexposed pixels


    #bad-border border defects, missing many pixel rows/columns on edges or showing digital info pixels on left edge. Also includes narrow or short border images.

    #bad-image one or more bad images in a set

    #bad-noise noise grain too large to make out many stars, obscures stars, pops up one-image 'stars'

    #bad-set too many bad images or bad noise to find asteroids. Bad set box marked


    #short-set set with less than 4 images, may still be usable (previously used #1-image-set)

    #long-set set with more than 4 images

    #25frames particular type of long set

    #5image-set particular type of long set


    #coma comatic aberration. Distortions, flat sides, 'horn' on many or most stars in an image. Mirror edge or framing defect

    #lens-flare centered circles, waves, may have spikes or gaps at cardinals, or 45s (also called #ghost)

    #lens-flare-spike especially if not in same image set as the likely lens-flare center


    #odd-set unusual questions with a set. May just be the first time I encounter a particular oddity

    Posted

  • MvGulik by MvGulik in response to AstroTinker's comment.

    Thanks for sharing that AstroTinker.

    Posted