import gradio as gr
from segment_anything import SamAutomaticMaskGenerator, sam_model_registry
import supervision as sv
from inference import DepthPredictor, SegmentPredictor
from utils import create_3d_obj
import numpy as np


def snap(image, video):
    depth_predictor = DepthPredictor()
    depth_result = depth_predictor.predict(image)
    gltf_path = create_3d_obj(np.array(image), depth_result)
    #segment_predictor = SegmentPredictor()
    #sam_result = segment_predictor.predict(image)
    return [gltf_path, video]#[depth_result, gltf_path, gltf_path]


demo = gr.Interface(
    snap,
    [gr.Image(source="webcam", tool=None, type="pil"), gr.Video(source="webcam")],
    ["image", "video"],
)

if __name__ == "__main__":
    demo.launch()