Spaces:
Running on Zero
Running on Zero
File size: 2,090 Bytes
7826761 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 | from __future__ import annotations
from pathlib import Path
import cv2
import numpy as np
import torch
from tqdm import tqdm
from camera import Camera
from tracker.inference.preprocess import enlarge_bbox
from .visualizer import Visualizer
def save_tracking_video(
output_path: str | Path,
video_input,
inference,
camera: Camera,
visualizer: Visualizer,
vertices: torch.Tensor,
fallback_fps: float,
) -> None:
fps = float(fallback_fps) if video_input.fps is None else float(video_input.fps)
video_input.reset()
frame_size = (int(inference.image_width * 2), int(inference.image_height * 2))
writer = cv2.VideoWriter(
str(output_path), cv2.VideoWriter_fourcc(*"mp4v"), fps, frame_size
)
if not writer.isOpened():
raise OSError(f"Failed to open video writer: {output_path}")
try:
num_frames = int(inference.vertex_coord.shape[0])
for frame_id in tqdm(range(num_frames), desc="visualize video"):
frame_rgb = video_input.read()
if frame_rgb is None:
raise ValueError(f"Unexpected end of input at frame {frame_id}.")
alignment = visualizer.vis_align(frame_rgb, inference.vertex_coord[frame_id])
alignment = visualizer.vis_contour(alignment, inference.vertex_coord[frame_id])
bbox = np.round(enlarge_bbox(inference.face_bbox[frame_id], 1.3)).astype(np.int32)
cv2.rectangle(alignment, bbox[:2], bbox[2:], (255, 0, 0), 2)
overlay, shading = visualizer.vis_flame_shading_overlay(
camera, frame_rgb, vertices[frame_id], alpha=0.5, return_shading=True
)
result = np.ascontiguousarray(
np.concatenate(
[
np.concatenate([frame_rgb, alignment], axis=1),
np.concatenate([overlay, shading], axis=1),
],
axis=0,
)
)
writer.write(cv2.cvtColor(result, cv2.COLOR_RGB2BGR))
finally:
writer.release()
|