app.py

import gradio as gr
import cv2
from PIL import Image
import numpy as np
from transformers import pipeline
import os
import torch
import torch.nn.functional as F
from torchvision import transforms
from torchvision.transforms import Compose
import trimesh
from geometry import create_triangles
import tempfile
from functools import partial
import spaces
from zipfile import ZipFile

from depth_anything.dpt import DepthAnything
from depth_anything.util.transform import Resize, NormalizeImage, PrepareForNet
from moviepy.editor import *

def zip_files(files_in, files_out):
    with ZipFile("depth_result.zip", "w") as zipObj:
        for idx, file in enumerate(files_in):
            zipObj.write(file, file.split("/")[-1])
        for idx, file in enumerate(files_out):
            zipObj.write(file, file.split("/")[-1])
    return "depth_result.zip"

def create_video(frames, fps, type):
    print("building video result")
    clip = ImageSequenceClip(frames, fps=fps)
    clip.write_videofile(type + "_result.mp4", fps=fps)
    
    return type + "_result.mp4"

@torch.no_grad()
def predict_depth(model, image):
    return model(image)["depth"]

@spaces.GPU
def make_video(video_path, outdir='./vis_video_depth', encoder='vits'):
    if encoder not in ["vitl","vitb","vits"]:
        encoder = "vits"

    mapper = {"vits":"small","vitb":"base","vitl":"large"}
    # DEVICE = 'cuda' if torch.cuda.is_available() else 'cpu'
    # model = DepthAnything.from_pretrained('LiheYoung/depth_anything_vitl14').to(DEVICE).eval()
    # Define path for temporary processed frames
    temp_frame_dir = tempfile.mkdtemp()
    
    margin_width = 50
    to_tensor_transform = transforms.ToTensor()

    DEVICE = 'cuda' if torch.cuda.is_available() else 'cpu'
    # depth_anything = DepthAnything.from_pretrained('LiheYoung/depth_anything_{}14'.format(encoder)).to(DEVICE).eval()
    depth_anything = pipeline(task = "depth-estimation", model=f"nielsr/depth-anything-{mapper[encoder]}")
    
    # total_params = sum(param.numel() for param in depth_anything.parameters())
    # print('Total parameters: {:.2f}M'.format(total_params / 1e6))
    
    transform = Compose([
        Resize(
            width=518,
            height=518,
            resize_target=False,
            keep_aspect_ratio=True,
            ensure_multiple_of=14,
            resize_method='lower_bound',
            image_interpolation_method=cv2.INTER_CUBIC,
        ),
        NormalizeImage(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
        PrepareForNet(),
    ])

    if os.path.isfile(video_path):
        if video_path.endswith('txt'):
            with open(video_path, 'r') as f:
                lines = f.read().splitlines()
        else:
            filenames = [video_path]
    else:
        filenames = os.listdir(video_path)
        filenames = [os.path.join(video_path, filename) for filename in filenames if not filename.startswith('.')]
        filenames.sort()
    
    # os.makedirs(outdir, exist_ok=True)
    
    for k, filename in enumerate(filenames):
        file_size = os.path.getsize(filename)/1024/1024
        if file_size > 128.0:
            print(f'File size of {filename} larger than 128Mb, sorry!')
            return filename
        print('Progress {:}/{:},'.format(k+1, len(filenames)), 'Processing', filename)
        
        raw_video = cv2.VideoCapture(filename)
        frame_width, frame_height = int(raw_video.get(cv2.CAP_PROP_FRAME_WIDTH)), int(raw_video.get(cv2.CAP_PROP_FRAME_HEIGHT))
        frame_rate = int(raw_video.get(cv2.CAP_PROP_FPS))
        if frame_rate < 1:
            frame_rate = 1
        cframes = int(raw_video.get(cv2.CAP_PROP_FRAME_COUNT))
        print(f'frames: {cframes}, fps: {frame_rate}')
        # output_width = frame_width * 2 + margin_width
        
        #filename = os.path.basename(filename)
        # output_path = os.path.join(outdir, filename[:filename.rfind('.')] + '_video_depth.mp4')
        #with tempfile.NamedTemporaryFile(delete=False, suffix='.mp4') as tmpfile:
        #    output_path = tmpfile.name
        #out = cv2.VideoWriter(output_path, cv2.VideoWriter_fourcc(*"avc1"), frame_rate, (output_width, frame_height))
        #fourcc = cv2.VideoWriter_fourcc(*'mp4v')
        #out = cv2.VideoWriter(output_path, fourcc, frame_rate, (output_width, frame_height))
        count=0
        depth_frames = []
        orig_frames = []
        while raw_video.isOpened():
            ret, raw_frame = raw_video.read()
            if not ret:
                break

            frame = cv2.cvtColor(raw_frame, cv2.COLOR_BGR2RGB) / 255.0
            frame_pil =  Image.fromarray((frame * 255).astype(np.uint8))
            frame = transform({'image': frame})['image']
            
            frame = torch.from_numpy(frame).unsqueeze(0).to(DEVICE)
            
            
            depth = to_tensor_transform(predict_depth(depth_anything, frame_pil))

            depth = F.interpolate(depth[None], (frame_height, frame_width), mode='bilinear', align_corners=False)[0, 0]
            depth = (depth - depth.min()) / (depth.max() - depth.min()) * 255.0
            
            depth = depth.cpu().numpy().astype(np.uint8)
            depth_color = cv2.applyColorMap(depth, cv2.COLORMAP_BONE)
            depth_gray = cv2.cvtColor(depth_color, cv2.COLOR_RGBA2GRAY)
            depth_color = cv2.cvtColor(depth_gray, cv2.COLOR_GRAY2RGB)

            # Remove white border around map:
            # define lower and upper limits of white
            white_lo = np.array([250,250,250])
            white_hi = np.array([255,255,255])
            # mask image to only select white
            mask = cv2.inRange(depth_color, white_lo, white_hi)
            # change image to black where we found white
            depth_color[mask>0] = (0,0,0)
            
            # split_region = np.ones((frame_height, margin_width, 3), dtype=np.uint8) * 255
            # combined_frame = cv2.hconcat([raw_frame, split_region, depth_color])
            
            # out.write(combined_frame)
            # frame_path = os.path.join(temp_frame_dir, f"frame_{count:05d}.png")
            # cv2.imwrite(frame_path, combined_frame)
            
            cv2.imwrite(f"f{count}.jpg", raw_frame)
            orig_frames.append(f"f{count}.jpg")
            cv2.imwrite(f"f{count}_dmap.jpg", depth_color)
            depth_frames.append(f"f{count}_dmap.jpg")
            count += 1

        final_vid = create_video(depth_frames, frame_rate, "depth")
        final_zip = zip_files(orig_frames, depth_frames)
        raw_video.release()
        # out.release()
        cv2.destroyAllWindows()
        
        return final_vid, final_zip, orig_frames[0], depth_frames[0] #output_path

def depth_edges_mask(depth):
    """Returns a mask of edges in the depth map.
    Args:
    depth: 2D numpy array of shape (H, W) with dtype float32.
    Returns:
    mask: 2D numpy array of shape (H, W) with dtype bool.
    """
    # Compute the x and y gradients of the depth map.
    depth_dx, depth_dy = np.gradient(depth)
    # Compute the gradient magnitude.
    depth_grad = np.sqrt(depth_dx ** 2 + depth_dy ** 2)
    # Compute the edge mask.
    mask = depth_grad > 0.05
    return mask

def pano_depth_to_world_points(depth):
    """
    360 depth to world points
    given 2D depth is an equirectangular projection of a spherical image
    Treat depth as radius
    longitude : -pi to pi
    latitude : -pi/2 to pi/2
    """

    # Convert depth to radius
    radius = 255 - depth.flatten()

    lon = np.linspace(0, np.pi*2, depth.shape[1])
    lat = np.linspace(0, np.pi, depth.shape[0])
    lon, lat = np.meshgrid(lon, lat)
    lon = lon.flatten()
    lat = lat.flatten()

    # Convert to cartesian coordinates
    x = radius * np.cos(lon) * np.sin(lat)
    y = radius * np.cos(lat)
    z = radius * np.sin(lon) * np.sin(lat)

    pts3d = np.stack([x, y, z], axis=1)
    uv = np.stack([lon, lat], axis=1)

    return [pts3d, uv]

def rgb2gray(rgb):
    return np.dot(rgb[...,:3], [0.333, 0.333, 0.333])

def get_mesh(image, depth):
    gdepth = rgb2gray(depth)
    print('depth to gray - ok')
    points = pano_depth_to_world_points(gdepth)
    pts3d = points[0]
    uv = points[1]
    print('radius from depth - ok')

    # Create a trimesh mesh from the points
    # Each pixel is connected to its 4 neighbors
    # colors are the RGB values of the image

    verts = pts3d.reshape(-1, 3)
    triangles = create_triangles(image.shape[0], image.shape[1])
    print('triangles - ok')
    colors = image.reshape(-1, 3)
    #mesh = trimesh.Trimesh(vertices=verts, faces=triangles, vertex_colors=colors)
    mesh = trimesh.PointCloud(verts, colors=colors)
    #material = trimesh.visual.texture.SimpleMaterial(image=image)
    #texture = trimesh.visual.TextureVisuals(uv=uv, image=image, material=material)
    #mesh.visual = texture
    scene = trimesh.Scene([mesh])
    print('mesh - ok')

    # Save as glb
    glb_file = tempfile.NamedTemporaryFile(suffix='.glb', delete=False)
    glb_path = glb_file.name
    scene.export(glb_path)
    print('file - ok')
    return glb_path

def loadurl(url):
    return url


css = """
#img-display-container {
    max-height: 100vh;
    }
#img-display-input {
    max-height: 80vh;
    }
#img-display-output {
    max-height: 80vh;
    }
"""


title = "# Depth Anything Video Demo"
description = """Depth Anything on full video files.  
Please refer to our [paper](https://arxiv.org/abs/2401.10891), [project page](https://depth-anything.github.io), or [github](https://github.com/LiheYoung/Depth-Anything) for more details.  
Mesh rendering from [ZoeDepth](https://huggingface.co/spaces/shariqfarooq/ZoeDepth) ([github](https://github.com/isl-org/ZoeDepth/tree/main/ui))."""

transform = Compose([
        Resize(
            width=518,
            height=518,
            resize_target=False,
            keep_aspect_ratio=True,
            ensure_multiple_of=14,
            resize_method='lower_bound',
            image_interpolation_method=cv2.INTER_CUBIC,
        ),
        NormalizeImage(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
        PrepareForNet(),
])

# @torch.no_grad()
# def predict_depth(model, image):
#     return model(image)

with gr.Blocks(css=css) as demo:
    gr.Markdown(title)
    gr.Markdown(description)
    gr.Markdown("### Video Depth Prediction demo")

    with gr.Row():
        with gr.Column():
            input_url = gr.Textbox(value="./examples/streetview.mp4", label="URL")
            input_video = gr.Video(label="Input Video", format="mp4")
            input_url.change(fn=loadurl, inputs=[input_url], outputs=[input_video])
            submit = gr.Button("Submit")
            output_frame = gr.Image(label="Frame", type='numpy')
            output_depth = gr.Image(label="Depth", type='numpy')
            render = gr.Button("Render")
        with gr.Column():
            model_type = gr.Dropdown([("small", "vits"), ("base", "vitb"), ("large", "vitl")], type="value", value="vits", label='Model Type')
            processed_video = gr.Video(label="Output Video", format="mp4")
            processed_zip = gr.File(label="Output Archive")
            result = gr.Model3D(label="3D Mesh", clear_color=[0.0, 0.0, 0.0, 0.0], camera_position=[0, 90, 0])
    
    def on_submit(uploaded_video,model_type):
                
        # Process the video and get the path of the output video
        output_video_path = make_video(uploaded_video,encoder=model_type)

        return output_video_path

    submit.click(on_submit, inputs=[input_video, model_type], outputs=[processed_video, processed_zip, output_frame, output_depth])
    render.click(partial(get_mesh), inputs=[output_frame, output_depth], outputs=[result])

    example_files = os.listdir('examples')
    example_files.sort()
    example_files = [os.path.join('examples', filename) for filename in example_files]
    examples = gr.Examples(examples=example_files, inputs=[input_video], outputs=[processed_video, processed_zip, output_frame, output_depth], fn=on_submit, cache_examples=True)
    

if __name__ == '__main__':
    demo.queue().launch()