GMU-Kitchens Scenes
===================
Contains 9 scenes {'gmu_scene_001', ..., 'gmu_scene_009'}

Each scene directory is named after the scene-name and has the following four subdirectories:
i)  Images: containing the rgb images of a scene

ii) Depths: containing the depth images of a scene

iii) Annotations: containing the xml files with the bounding box annotations of just the 11 BigBird objects.
Each annotation file has the same name as the rgb image in the 'Images' subdirectory.
We followed the same format as the VOC2007/VOC2012 datasets.

iv) Annotations_withExtra: similar to Annotations but the xml files in this folder also contain ground truth bounding boxes for the extra objects that can be found scattered in the scenes.


Object categories and corresponding labels
==========================================
	BigBird Objects:
coca_cola_glass_bottle 1
coffee_mate_french_vanilla 2
honey_bunches_of_oats_honey_roasted 3
hunts_sauce 4
mahatma_rice 5
nature_valley_soft_baked_oatmeal_squares_cinnamon_brown_sugar 6
nature_valley_sweet_and_salty_nut_almond 7
palmolive_orange 8
pop_secret_light_butter 9
pringles_bbq 10
red_bull 11

	Extra Objects (and scenes they appear)
bowl_1 	12  	scenes 1,2
bowl_2 	13  	scenes 1,2
bowl_3	14	scenes 5,6
bowl_4	15	scenes 5,6
bowl_5	16	scene 9
mug_1  	17  	scenes 1,2
mug_2   18  	scenes 1,2
mug_3   19	scenes 5,6
mug_4	20	scenes 5,6
mug_5	21	scene 9
salt    22  	scenes 1,2
cleaner 23      scenes 3,4
 

Train/test split:
=================
We did a 3-fold cross validated experiment. We did a random train/test split according to the scene names and prepared the following folds:

FOLD 1:
-------
train fold:{gmu_scene_001, gmu_scene_003, gmu_scene_006, gmu_scene_007, gmu_scene_008, gmu_scene_009} 
test  fold:{gmu_scene_002, gmu_scene_004, gmu_scene_005}

FOLD 2:
-------
train fold:{gmu_scene_001, gmu_scene_002, gmu_scene_004, gmu_scene_005, gmu_scene_007, gmu_scene_008} 
test  fold:{gmu_scene_003, gmu_scene_006, gmu_scene_009}

FOLD 3:
-------
train fold:{gmu_scene_002, gmu_scene_003, gmu_scene_004, gmu_scene_005, gmu_scene_006, gmu_scene_009} 
test  fold:{gmu_scene_001, gmu_scene_007, gmu_scene_008}


For each fold, we trained a model using images of 6 scenes in the train fold.