video-intelligence / labels /default.json
Zmu's picture
Split backend into hub.py, detection.py, captioning.py; zero-shot labels from a JSON vocabulary (labels/default.json) with typed override
b5a226e verified
Raw History Blame Contribute Delete
4.02 kB
{
"person": ["a person", "a cartoon character", "an animated character"],
"bicycle": ["a bicycle", "a cartoon bicycle"],
"car": ["a car", "a cartoon car"],
"motorcycle": ["a motorcycle", "a cartoon motorcycle"],
"airplane": ["an airplane", "a cartoon airplane"],
"bus": ["a bus", "a cartoon bus"],
"train": ["a train", "a cartoon train"],
"truck": ["a truck", "a cartoon truck"],
"boat": ["a boat", "a cartoon boat"],
"traffic light": ["a traffic light"],
"fire hydrant": ["a fire hydrant", "a cartoon fire hydrant"],
"stop sign": ["a stop sign"],
"parking meter": ["a parking meter"],
"bench": ["a bench", "a park bench"],
"bird": ["a bird", "a cartoon bird"],
"cat": ["a cat", "a cartoon cat"],
"dog": ["a dog", "a cartoon dog"],
"horse": ["a horse", "a cartoon horse"],
"sheep": ["a sheep", "a cartoon sheep"],
"cow": ["a cow", "a cartoon cow"],
"elephant": ["an elephant", "a cartoon elephant"],
"bear": ["a bear", "a cartoon bear"],
"zebra": ["a zebra", "a cartoon zebra"],
"giraffe": ["a giraffe", "a cartoon giraffe"],
"backpack": ["a backpack", "a cartoon backpack"],
"umbrella": ["an umbrella", "a cartoon umbrella"],
"handbag": ["a handbag", "a purse"],
"tie": ["a necktie", "a tie"],
"suitcase": ["a suitcase", "luggage"],
"frisbee": ["a frisbee"],
"skis": ["skis", "a pair of skis"],
"snowboard": ["a snowboard"],
"sports ball": ["a ball", "a sports ball"],
"kite": ["a kite"],
"baseball bat": ["a baseball bat"],
"baseball glove": ["a baseball glove", "a baseball mitt"],
"skateboard": ["a skateboard", "a cartoon skateboard"],
"surfboard": ["a surfboard"],
"tennis racket": ["a tennis racket"],
"bottle": ["a bottle", "a cartoon bottle"],
"wine glass": ["a wine glass"],
"cup": ["a cup", "a mug", "a cartoon cup"],
"fork": ["a fork"],
"knife": ["a knife"],
"spoon": ["a spoon"],
"bowl": ["a bowl"],
"banana": ["a banana", "a cartoon banana"],
"apple": ["an apple", "a cartoon apple"],
"sandwich": ["a sandwich", "a cartoon sandwich"],
"orange": ["an orange"],
"broccoli": ["broccoli"],
"carrot": ["a carrot", "a cartoon carrot"],
"hot dog": ["a hot dog"],
"pizza": ["a pizza", "a slice of pizza"],
"donut": ["a donut", "a cartoon donut"],
"cake": ["a cake", "a cartoon cake"],
"chair": ["a chair", "a cartoon chair"],
"couch": ["a couch", "a sofa", "a cartoon couch"],
"potted plant": ["a potted plant", "a houseplant"],
"bed": ["a bed", "a cartoon bed"],
"dining table": ["a dining table", "a table"],
"toilet": ["a toilet", "a cartoon toilet"],
"tv": ["a television", "a tv monitor", "a cartoon tv"],
"laptop": ["a laptop computer", "a cartoon laptop"],
"mouse": ["a computer mouse"],
"remote": ["a remote control"],
"keyboard": ["a computer keyboard"],
"cell phone": ["a cell phone", "a smartphone"],
"microwave": ["a microwave oven"],
"oven": ["an oven", "a kitchen stove"],
"toaster": ["a toaster"],
"sink": ["a kitchen sink", "a bathroom sink"],
"refrigerator": ["a refrigerator", "a cartoon fridge"],
"book": ["a book", "a cartoon book"],
"clock": ["a clock", "a wall clock", "a cartoon clock"],
"vase": ["a vase"],
"scissors": ["a pair of scissors"],
"teddy bear": ["a teddy bear", "a stuffed animal"],
"hair drier": ["a hair dryer"],
"toothbrush": ["a toothbrush"],
"gun": ["a handgun", "a pistol", "a rifle", "a cartoon gun"],
"sword": ["a sword", "a blade"],
"door": ["a door", "a doorway"],
"window": ["a window"],
"lamp": ["a table lamp", "a floor lamp"],
"mirror": ["a mirror"],
"plate": ["a plate", "a dinner plate"],
"box": ["a cardboard box", "a shipping box"],
"guitar": ["a guitar", "an electric guitar"],
"hat": ["a hat", "a cap", "a cartoon hat"],
"glasses": ["eyeglasses", "sunglasses"],
"rabbit": ["a rabbit", "a cartoon rabbit", "a bunny"],
"duck": ["a duck", "a cartoon duck"],
"rodent": ["a mouse", "a rat", "a cartoon rodent"],
"robot": ["a robot", "a cartoon robot", "a mechanical droid"]
}