Is the fine-grained action label annotated in this dataset?

[vanoracai]

Thanks for sharing your great work! I notice that "Each of those action categories is the collection of many different and specific actions. E.g. Gaming includes Boxing, Shooting Gun, Playing Golf, Playing Baseball just to cite a few."

I run the demo.py file and it seems the action label is within the nine action categories mentioned in the table. May I ask if there's any fine-grained action label for each frame? say, boxing instead of gaming?

[DenisTome]

In the file utils/config.py you can see the ACTION variable containing the mapping from the fine action categories to the broader one displayed in the evaluation matrices in the paper.

ACTION = {
        'anim_Clip1': 8, 'Opening_A_Lid': 0, 'Dribble': 5, 'Boxing': 5,
        'Standing_Arguing__1_': 3, 'Happy': 3, 'Plotting': 3, 'Counting': 4,
        'Standing_Arguing': 0, 'Standing_2H_Cast_Spell_01': 4, 'Shooting_Gun': 5,
        'Two_Hand_Spell_Casting': 0, 'Shaking_Hands_2': 2, 'Hands_Forward_Gesture': 2,
        'Rifle_Punch': 1, 'Baseball_Umpire': 5, 'Angry_Gesture': 0, 'Waving_Gesture': 0,
        'Taunt_Gesture': 0, 'Golf_Putt_Failure': 5, 'Rejected': 1, 'Shake_Fist': 2,
        'Revealing_Dice': 5, 'Golf_Putt_Failure__1_': 5, 'No': 3, 'Angry_Point': 1,
        'Agreeing': 3, 'Sitting_Thumbs_Up': 6, 'Standing_Thumbs_Up': 4, 'Patting': 7,
        'Petting': 7, 'Petting_Animal': 7, 'Taking_Punch': 0,
        'Standing_1H_Magic_Attack_01': 4, 'Talking': 3, 'Standing_Greeting': 2,
        'Happy_Hand_Gesture': 0, 'Dismissing_Gesture': 1, 'Strong_Gesture': 1,
        'Pointing_Gesture': 1, 'Golf_Putt_Victory': 5, 'Pointing': 0,
        'Thinking': 4, 'Loser': 1, 'Reaching_Out': 3, 'Crazy_Gesture': 0,
        'Golf_Putt_Victory__1_': 5, 'Insult': 3, 'Arm_Gesture': 0,
        'Beckoning': 1, 'Charge': 5, 'Weight_Shift_Gesture': 8,
        'Pain_Gesture': 1, 'Fist_Pump': 0, 'Terrified': 1, 'Surprised': 1,
        'Clapping': 1, 'Rallying': 1, 'Hand_Raising': 0, 'Sitting_Disapproval': 6,
        'Quick_Formal_Bow': 2, 'Counting__1_': 0, 'Tpose_Take_001': 4,
        'upper_stretching': 4, 'lower_stretching': 6, 'walking': 8
    }

for each element in that dictionary, the key is mapped to the index of the broader action name coming from

_NAMES = [
        'Gesticuling', 'Reacting', 'Greeting',
        'Talking', 'UpperStretching', 'Gaming',
        'LowerStretching', 'Patting', 'Walking', 'All'
    ]

If you want to assess your results on the individual finer actions, just don't apply this mapping.

[vanoracai]

Hi Denis, thanks for your kind answer!