|
The annotation files are divided into several parts, where only two are of interrest in this case. The first part is the #Action_Gesture_List which describes the numbers given to each gesture, as shown below. |
#PATH ./ #TYPE bmp #Person_Modality_List 00 left #ENDLIST #Action_Gesture_List 00 hand 01 point 02 point_thumb 03 uncertain #ENDLIST Layer Name Type Modality Visible #Annotation_Layers 00 left 3 0 1 #ENDLIST Object format: object imagenumber layer no_of_contours Contour format: contournumber inner no_of_points mp1 mp2 [x y]*no_points #OBJECTLIST #ENDLIST #FINGERLIST #ENDLIST |
|
The second part is the actual annotation of the sequences. Here only 3 parts of each line is of interrest : The Gesture, the Action_Start, and the Gesture_End. The Gesture column contains the numbers given to the gestures in the first part, the Action_Start contains the name of the first picture of the gesture, and the Gesture_End the name of the last picture. |
Event Layer Person Gesture Action_Start Stroke_Start Stroke_End Gesture_End #EVENTLIST 0000 00 00 03 X_01_W_M_0118.bmp X_01_W_M_0118.bmp X_01_W_M_0121.bmp X_01_W_M_0121.bmp 0001 00 00 02 X_01_W_M_0122.bmp X_01_W_M_0122.bmp X_01_W_M_0143.bmp X_01_W_M_0143.bmp 0002 00 00 01 X_01_W_M_0144.bmp X_01_W_M_0144.bmp X_01_W_M_0149.bmp X_01_W_M_0149.bmp |
|
The reason the annotation files include more information than we need, is that the annotation prorgam can have more features than used in this case. |