app.py

from main import run_app, run_train, run_inference

import spaces
from PIL import Image
import cv2
import os
import gradio as gr

with gr.Blocks() as demo:
    with gr.Row():
        with gr.Column():
            char_imgs = gr.Gallery(type="pil", label="Images of the Character")
            mocap = gr.Video(label="Motion-Capture Video")
            tr_steps = gr.Number(label="Training steps", value=10)
            inf_steps = gr.Number(label="Inference steps", value=10)
            fps = gr.Number(label="Output frame rate", value=12)
            modelId = gr.Text(label="Model Id", value="fine_tuned_pcdms")
            remove_bg = gr.Checkbox(label="Remove background", value=False)
            resize_inputs = gr.Checkbox(label="Resize images to match video", value=True)
            img_width = gr.Number(label="Output width", value=1920)
            img_height = gr.Number(label="Output height", value=1080)
            train_btn = gr.Button(value="Train")
            inference_btn = gr.Button(value="Inference")
            submit_btn = gr.Button(value="Generate")
        with gr.Column():
            animation = gr.Video(label="Result")
            frames = gr.Gallery(type="pil", label="Frames", format="png")
            frames_thumb = gr.Gallery(type="pil", label="Thumbnails", format="png")
            pose_coords = gr.JSON(label="Pose Coordinates")

    submit_btn.click(
        run_app, inputs=[char_imgs, mocap, tr_steps, inf_steps, fps, remove_bg, resize_inputs], outputs=[animation, frames]
    )
    
    train_btn.click(
        run_train, inputs=[char_imgs, tr_steps, modelId, remove_bg, resize_inputs], outputs=[]
    )
    
    inference_btn.click(
        run_inference, inputs=[char_imgs, mocap, tr_steps, inf_steps, fps, modelId, img_width, img_height, remove_bg, resize_inputs], outputs=[animation, frames, frames_thumb, pose_coords]
    )


demo.launch(share=True)