You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In an inference procedure, when we try to use get_view method to show the inference, the inference is ignored if we push it directly inside the method, drawing the GT objects and thus generates 'false views'. The following code try to explain better the idea:
import alonet
detr = alonet.detr.LitDetr(weights = "detr-r50")
data_loader = alonet.detr.CocoDetection2Detr(sample = True, batch_size = 1)
frame = next(iter(data_loader.val_dataloader()))
frame = frame[0].batch_list(frame)
pred_boxes = detr.inference(detr(frame))
pred_boxes[0].labels[:] = 1000 # Change ID to explain better the problem
# Ignore predictions and draw GT
frame.get_view([pred_boxes[0]]).render()
# Correct view
frame.get_view([pred_boxes[0].get_view(frame[0])]).render()
Some of this things to do:
Launch an error to show the misunderstanding of the use of aloscene.
Update get_view to enable this feature.
The text was updated successfully, but these errors were encountered:
In an inference procedure, when we try to use
get_view
method to show the inference, the inference is ignored if we push it directly inside the method, drawing the GT objects and thus generates 'false views'. The following code try to explain better the idea:Some of this things to do:
The text was updated successfully, but these errors were encountered: