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
I have an issue while converting MOTS(Multiple object tracking segmentation) ground truth values to standard MOT(Multiple object tracking) formats.
MOTS Format
For example:1 2002 2 1080 1920 UkU\1`0RQ1>PoN\OVP1X1F=I3oSOTNlg0U2lWOVNng0m1nWOWNlg0n1PXOWNlg0l1SXOUNjg0P2RXO....
Frame: 1,
class 2(person), index of the person: 002
class_id: 2 (person)
height: 1080 width: 1920 of the image
RLE: UkU\1`0RQ1>PoN\OVP1X1F=I3oSOTNlg0U2lWOVNng0m1nWOWNlg0n1PXOWNlg0l1SXOUNjg0P2RXO....
MOT format with Bounding boxes
example: 1 1 78.2 47.3 71.5 178.6 1 1 0.8
Frame: 1
identity no 1
Bounding box left 78.2
Bounding box top 47.3
Bounding box width 71.5
Bounding box height 178.6
confidence score 1: It acts as a flag whether the entry is to be considered (1) or ignored (0)
8.Class 1 (here 1 is for person): Indicates the type of object annotated
9.Visibility: 0.8 : Visibility ratio, a number between 0 and 1 that says how much of that object is visible. Can be due to occlusion and due to image border cropping.
How to convert RLE to the above data and I can send the gt files as well if required?
All the data is present in txt files
The text was updated successfully, but these errors were encountered:
I have an issue while converting MOTS(Multiple object tracking segmentation) ground truth values to standard MOT(Multiple object tracking) formats.
MOTS Format
For example:1 2002 2 1080 1920 UkU\1`0RQ1>PoN\OVP1X1F=I3oSOTNlg0U2lWOVNng0m1nWOWNlg0n1PXOWNlg0l1SXOUNjg0P2RXO....
Frame: 1,
class 2(person), index of the person: 002
class_id: 2 (person)
height: 1080 width: 1920 of the image
RLE: UkU\1`0RQ1>PoN\OVP1X1F=I3oSOTNlg0U2lWOVNng0m1nWOWNlg0n1PXOWNlg0l1SXOUNjg0P2RXO....
MOT format with Bounding boxes
example: 1 1 78.2 47.3 71.5 178.6 1 1 0.8
Frame: 1
identity no 1
Bounding box left 78.2
Bounding box top 47.3
Bounding box width 71.5
Bounding box height 178.6
confidence score 1: It acts as a flag whether the entry is to be considered (1) or ignored (0)
8.Class 1 (here 1 is for person): Indicates the type of object annotated
9.Visibility: 0.8 : Visibility ratio, a number between 0 and 1 that says how much of that object is visible. Can be due to occlusion and due to image border cropping.
How to convert RLE to the above data and I can send the gt files as well if required?
All the data is present in txt files
The text was updated successfully, but these errors were encountered: