Closed ltttpku closed 1 year ago
Hi @ltttpku,
As the name suggests, object_to_action
shows the correspondence between the objects and actions. So, the key 1
is the object index and the list [0, 3, 11, 15]
is the valid actions for that object.
It does seem a bit weird to have person
as the potential object for cut object
. But this list of correspondence was generated from the training data, which means there must be training examples for that particular combination. I will need to take a look at the dataset later.
Fred.
Hi @ltttpku,
I have checked the dataset, and can confirm that all those actions do exist.
In particular, the cutting obj
and person
pair can happen in a surgery scene or barber's shop. You can use the dataset navigator utility as provided in the repo to check these images. The image indices are 559
, 2173
, 2363
and 4416
.
Fred.
That makes sense! Closing the issue :p
There're some cool properties of VCOCO dataset you implemented: "
object_to_action
" gives me the list of actions for each object, i.e.{1: [0, 3, 11, 15], 2: [0, 1, 2, 3, 11], ......}
"objects
" return the list of objects, i.e.['background', 'person', 'bicycle', .......]
"actions
" return the list of actions, i.e.['hold obj', 'sit instr', 'ride instr', .......]
However, I'm confused about the relationships among them:
1
of "1: [0, 3, 11, 15]
", which is the first item of object_to_action, represent?[0, 3, 11, 15]
of "1: [0, 3, 11, 15]
" represent?According to the List of actions and objects,
Actions 0, 3, 11, 15
representhold obj, look obj, carry obj, cut obj
respectively whileObject 1
representperson
, which appears to be weird.