Datasets:

xingzhaohu
/

syncnet_compute

License:

syncnet_compute / ConstructDataJson /1_construct_video_path.py

Add files using upload-large-folder tool

d26d563 verified 8 months ago

9.97 kB

import os
import json
import cv2
from typing import List, Optional, Sequence, Tuple
from collections import Counter
import multiprocessing
from tqdm import tqdm  # 导入tqdm
import numpy as np
import ffmpeg


def _read_labels_from_video(video_path: str) -> Optional[np.ndarray]:
    """Read grayscale label video back as numpy array: (T, H, W), uint8."""
    try:
        probe = ffmpeg.probe(video_path)
        video_info = next(s for s in probe["streams"] if s["codec_type"] == "video")
        width = int(video_info["width"])
        height = int(video_info["height"])

        out, _ = (
            ffmpeg.input(video_path)
            .output("pipe:", format="rawvideo", pix_fmt="gray")
            .run(capture_stdout=True, capture_stderr=True)
        )

        decoded = np.frombuffer(out, np.uint8).reshape((-1, height, width))
        return decoded
    except Exception as e:
        print(f"Error reading label video {video_path}: {e}")
        return None


def _compute_lip_bboxes(
    labels: np.ndarray,
    lip_scale: float = 1.0,
    nose_labels: Sequence[int] = (2,),
    face_labels: Sequence[int] = (1,),
) -> List[Optional[Tuple[int, int, int, int]]]:
    """Compute per-frame mouth-region bboxes using nose + face masks, with temporal interpolation."""
    if labels.ndim != 3:
        raise ValueError("labels must have shape (T, H, W)")

    T, H, W = labels.shape
    lip_scale = max(float(lip_scale), 1.0)

    raw_bboxes: List[Optional[Tuple[int, int, int, int]]] = [None] * T

    for t in range(T):
        frame_labels = labels[t]

        nose_mask = np.isin(frame_labels, nose_labels)
        face_mask = np.isin(frame_labels, face_labels)

        if not np.any(nose_mask) or not np.any(face_mask):
            continue

        nose_ys, _ = np.where(nose_mask)
        y_top = float(nose_ys.max())

        face_ys, face_xs = np.where(face_mask)
        y_bottom = float(face_ys.max())
        x_left = float(face_xs.min())
        x_right = float(face_xs.max())

        if y_bottom <= y_top:
            continue

        x_min = x_left
        x_max = x_right
        y_min = y_top
        y_max = y_bottom

        w = x_max - x_min + 1.0
        h = y_max - y_min + 1.0
        cx = (x_min + x_max) / 2.0
        cy = (y_min + y_max) / 2.0

        new_w = w * lip_scale
        new_h = h * lip_scale

        x_min_s = int(round(cx - new_w / 2.0))
        x_max_s = int(round(cx + new_w / 2.0))
        y_min_s = int(round(cy - new_h / 2.0))
        y_max_s = int(round(cy + new_h / 2.0))

        x_min_s = max(0, min(x_min_s, W - 1))
        x_max_s = max(0, min(x_max_s, W - 1))
        y_min_s = max(0, min(y_min_s, H - 1))
        y_max_s = max(0, min(y_max_s, H - 1))

        if x_max_s <= x_min_s or y_max_s <= y_min_s:
            continue

        raw_bboxes[t] = (x_min_s, y_min_s, x_max_s, y_max_s)

    if not any(bb is not None for bb in raw_bboxes):
        return raw_bboxes

    coords: List[List[Optional[int]]] = [[None] * T for _ in range(4)]
    for t, bb in enumerate(raw_bboxes):
        if bb is None:
            continue
        for d in range(4):
            coords[d][t] = bb[d]

    for d in range(4):
        keyframes = [(t, coords[d][t]) for t in range(T) if coords[d][t] is not None]
        if not keyframes:
            continue

        first_idx, first_val = keyframes[0]
        for t in range(0, first_idx):
            coords[d][t] = first_val

        for (i, v0), (j, v1) in zip(keyframes, keyframes[1:]):
            coords[d][i] = v0
            coords[d][j] = v1
            gap = j - i
            if gap <= 1:
                continue
            for t in range(i + 1, j):
                alpha = (t - i) / float(gap)
                interp_val = int(round(v0 + (v1 - v0) * alpha))
                coords[d][t] = interp_val

        last_idx, last_val = keyframes[-1]
        for t in range(last_idx + 1, T):
            coords[d][t] = last_val

    final_bboxes: List[Optional[Tuple[int, int, int, int]]] = [None] * T
    for t in range(T):
        if all(coords[d][t] is not None for d in range(4)):
            final_bboxes[t] = (
                int(coords[0][t]),
                int(coords[1][t]),
                int(coords[2][t]),
                int(coords[3][t]),
            )

    return final_bboxes


def _bbox_area(bb: Tuple[int, int, int, int]) -> int:
    x_min, y_min, x_max, y_max = bb
    return max(0, x_max - x_min + 1) * max(0, y_max - y_min + 1)


def _has_area_jump(
    bboxes: List[Optional[Tuple[int, int, int, int]]],
    area_ratio_thresh: float = 1.5,
) -> bool:
    if area_ratio_thresh <= 1.0:
        return False

    prev_area: Optional[int] = None
    for bb in bboxes:
        if bb is None:
            continue
        area = _bbox_area(bb)
        if area <= 0:
            continue
        if prev_area is not None:
            ratio = max(area / prev_area, prev_area / area)
            if ratio >= area_ratio_thresh:
                return True
        prev_area = area
    return False

def find_mp4_files(directory):
    # 用于记录所有的视频路径
    video_paths = []

    # 遍历目录及子目录
    for root, dirs, files in os.walk(directory):
        for file in files:
            # 检查文件扩展名是否为.mp4
            if file.lower().endswith('.mp4'):
                # 获取视频的绝对路径
                video_paths.append(os.path.join(root, file))
    
    return video_paths

def get_frame_count(file_path):
    # 使用OpenCV读取视频并获取帧数
    cap = cv2.VideoCapture(file_path)

    if not cap.isOpened():
        return 0
    # 获取视频总帧数
    frame_count = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
    # fps = cap.get(cv2.CAP_PROP_FPS)                        # 帧率

    # print(f"fps is {fps}")
    cap.release()
    return frame_count

def process_video(video_path):
    # 处理每个视频，返回路径和帧数
    frame_count = get_frame_count(video_path)
    # if fps < 29 or fps > 31:
    global data_type
    label_path = None
    if data_type == "hallo3":
        label_path = video_path.replace("/videos/", "/face_parse_labels/").replace(".mp4", ".mp4.mkv")
    elif data_type == "MEAD":
        label_path = video_path.replace("/MEAD/", "/MEAD_face_labels/").replace(".mp4", ".mkv")
    elif data_type == "TED":
        label_path = video_path.replace("/test_clips_tian_scenecut_resampled/", "/test_clips_tian_scenecut_face_labels_resampled_face_alignment/").replace(".mp4", ".mkv")

    if label_path and os.path.exists(label_path):
        labels = _read_labels_from_video(label_path)
        if labels is None:
            return None
        bboxes = _compute_lip_bboxes(labels)
        if _has_area_jump(bboxes, area_ratio_thresh=1.5):
            return None
        return {"video_path": video_path, "frame_count": frame_count, "label_path": label_path}

    return None 
        # return {"video_path": video_path, "frame_count": frame_count, "label_path": video_path.replace("/test_clips_tian_scenecut_resampled/", "/test_clips_tian_scenecut_face_labels_resampled/").replace(".mp4", ".mkv")}

def save_to_jsonl(results, output_file):
    # 保存处理结果到JSONL文件
    with open(output_file, 'w', encoding='utf-8') as f:
        for result in results:
            f.write(json.dumps(result, ensure_ascii=False) + '\n')

def print_frame_distribution(frame_counts):
    # 使用Counter统计帧数分布
    frame_count_distribution = Counter(frame_counts)
    # fps_distribution = Counter(fps_counts)
    
    print("帧数分布情况：")
    for frame_count, count in sorted(frame_count_distribution.items()):
        print(f"帧数: {frame_count} - 文件数: {count}")

    # print("帧数分布情况：")
    # for fps_count, count in sorted(fps_distribution.items()):
    #     print(f"帧数: {fps_count} - 文件数: {count}")

def format_count_suffix(count):
    if count >= 1000:
        suffix = f"{count / 1000:.1f}k"
        return suffix.replace(".0k", "k")
    return str(count)

def main(directory, output_file):
    # 步骤1：获取所有mp4文件路径
    video_paths = find_mp4_files(directory)

    # 步骤2：使用多进程并行处理视频文件，并添加进度条
    with multiprocessing.Pool() as pool:
        # 使用tqdm结合multiprocessing.Pool.map，显示进度条
        results = list(tqdm(pool.imap(process_video, video_paths), total=len(video_paths), desc="处理视频"))

    filter_results = [res for res in results if res is not None and res['frame_count'] > 90]

    # 步骤3：提取所有的帧数并打印帧数分布情况
    frame_counts = [result['frame_count'] for result in filter_results]
    # fps_counts = [round(result['fps']) for result in filter_results]
    print_frame_distribution(frame_counts)

    # 步骤4：保存结果到JSONL文件
    count_suffix = format_count_suffix(len(filter_results))
    base, ext = os.path.splitext(output_file)
    output_file_with_count = f"{base}_{count_suffix}{ext}"
    save_to_jsonl(filter_results, output_file_with_count)

    print(f"len results is {len(filter_results)}")
    print(f"output file is {output_file_with_count}")

    print("完成！所有的.mp4文件路径和帧数已经保存为 JSONL 格式。")

# # 示例调用
# directory = '/share/zhaohu_workspace/Downloaded_Data/MEAD'  # 替换为目标目录路径
# # output_file = 'MEAD_f90_facealignment.jsonl'  # 输出文件名
# output_file = 'MEAD_f90.jsonl'  # 输出文件名
# data_type = "MEAD"


# 示例调用
directory = '/share/zhaohu_workspace/Downloaded_Data/hallo3_training_data/videos'  # 替换为目标目录路径
output_file = 'hallo3_f90.jsonl'  # 输出文件名
data_type = "hallo3"


# directory = '/share/zhaohu_workspace/Downloaded_Data/ted_clips/test_clips_tian_scenecut_resampled'  # 替换为目标目录路径
# data_type = "TED"
# output_file = 'TED_f90_facealignment.jsonl'  # 输出文件名

main(directory, output_file)

	import os
	import json
	import cv2
	from typing import List, Optional, Sequence, Tuple
	from collections import Counter
	import multiprocessing
	from tqdm import tqdm # 导入tqdm
	import numpy as np
	import ffmpeg


	def _read_labels_from_video(video_path: str) -> Optional[np.ndarray]:
	"""Read grayscale label video back as numpy array: (T, H, W), uint8."""
	try:
	probe = ffmpeg.probe(video_path)
	video_info = next(s for s in probe["streams"] if s["codec_type"] == "video")
	width = int(video_info["width"])
	height = int(video_info["height"])

	out, _ = (
	ffmpeg.input(video_path)
	.output("pipe:", format="rawvideo", pix_fmt="gray")
	.run(capture_stdout=True, capture_stderr=True)
	)

	decoded = np.frombuffer(out, np.uint8).reshape((-1, height, width))
	return decoded
	except Exception as e:
	print(f"Error reading label video {video_path}: {e}")
	return None


	def _compute_lip_bboxes(
	labels: np.ndarray,
	lip_scale: float = 1.0,
	nose_labels: Sequence[int] = (2,),
	face_labels: Sequence[int] = (1,),
	) -> List[Optional[Tuple[int, int, int, int]]]:
	"""Compute per-frame mouth-region bboxes using nose + face masks, with temporal interpolation."""
	if labels.ndim != 3:
	raise ValueError("labels must have shape (T, H, W)")

	T, H, W = labels.shape
	lip_scale = max(float(lip_scale), 1.0)

	raw_bboxes: List[Optional[Tuple[int, int, int, int]]] = [None] * T

	for t in range(T):
	frame_labels = labels[t]

	nose_mask = np.isin(frame_labels, nose_labels)
	face_mask = np.isin(frame_labels, face_labels)

	if not np.any(nose_mask) or not np.any(face_mask):
	continue

	nose_ys, _ = np.where(nose_mask)
	y_top = float(nose_ys.max())

	face_ys, face_xs = np.where(face_mask)
	y_bottom = float(face_ys.max())
	x_left = float(face_xs.min())
	x_right = float(face_xs.max())

	if y_bottom <= y_top:
	continue

	x_min = x_left
	x_max = x_right
	y_min = y_top
	y_max = y_bottom

	w = x_max - x_min + 1.0
	h = y_max - y_min + 1.0
	cx = (x_min + x_max) / 2.0
	cy = (y_min + y_max) / 2.0

	new_w = w * lip_scale
	new_h = h * lip_scale

	x_min_s = int(round(cx - new_w / 2.0))
	x_max_s = int(round(cx + new_w / 2.0))
	y_min_s = int(round(cy - new_h / 2.0))
	y_max_s = int(round(cy + new_h / 2.0))

	x_min_s = max(0, min(x_min_s, W - 1))
	x_max_s = max(0, min(x_max_s, W - 1))
	y_min_s = max(0, min(y_min_s, H - 1))
	y_max_s = max(0, min(y_max_s, H - 1))

	if x_max_s <= x_min_s or y_max_s <= y_min_s:
	continue

	raw_bboxes[t] = (x_min_s, y_min_s, x_max_s, y_max_s)

	if not any(bb is not None for bb in raw_bboxes):
	return raw_bboxes

	coords: List[List[Optional[int]]] = [[None] * T for _ in range(4)]
	for t, bb in enumerate(raw_bboxes):
	if bb is None:
	continue
	for d in range(4):
	coords[d][t] = bb[d]

	for d in range(4):
	keyframes = [(t, coords[d][t]) for t in range(T) if coords[d][t] is not None]
	if not keyframes:
	continue

	first_idx, first_val = keyframes[0]
	for t in range(0, first_idx):
	coords[d][t] = first_val

	for (i, v0), (j, v1) in zip(keyframes, keyframes[1:]):
	coords[d][i] = v0
	coords[d][j] = v1
	gap = j - i
	if gap <= 1:
	continue
	for t in range(i + 1, j):
	alpha = (t - i) / float(gap)
	interp_val = int(round(v0 + (v1 - v0) * alpha))
	coords[d][t] = interp_val

	last_idx, last_val = keyframes[-1]
	for t in range(last_idx + 1, T):
	coords[d][t] = last_val

	final_bboxes: List[Optional[Tuple[int, int, int, int]]] = [None] * T
	for t in range(T):
	if all(coords[d][t] is not None for d in range(4)):
	final_bboxes[t] = (
	int(coords[0][t]),
	int(coords[1][t]),
	int(coords[2][t]),
	int(coords[3][t]),
	)

	return final_bboxes


	def _bbox_area(bb: Tuple[int, int, int, int]) -> int:
	x_min, y_min, x_max, y_max = bb
	return max(0, x_max - x_min + 1) * max(0, y_max - y_min + 1)


	def _has_area_jump(
	bboxes: List[Optional[Tuple[int, int, int, int]]],
	area_ratio_thresh: float = 1.5,
	) -> bool:
	if area_ratio_thresh <= 1.0:
	return False

	prev_area: Optional[int] = None
	for bb in bboxes:
	if bb is None:
	continue
	area = _bbox_area(bb)
	if area <= 0:
	continue
	if prev_area is not None:
	ratio = max(area / prev_area, prev_area / area)
	if ratio >= area_ratio_thresh:
	return True
	prev_area = area
	return False

	def find_mp4_files(directory):
	# 用于记录所有的视频路径
	video_paths = []

	# 遍历目录及子目录
	for root, dirs, files in os.walk(directory):
	for file in files:
	# 检查文件扩展名是否为.mp4
	if file.lower().endswith('.mp4'):
	# 获取视频的绝对路径
	video_paths.append(os.path.join(root, file))

	return video_paths

	def get_frame_count(file_path):
	# 使用OpenCV读取视频并获取帧数
	cap = cv2.VideoCapture(file_path)

	if not cap.isOpened():
	return 0
	# 获取视频总帧数
	frame_count = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
	# fps = cap.get(cv2.CAP_PROP_FPS) # 帧率

	# print(f"fps is {fps}")
	cap.release()
	return frame_count

	def process_video(video_path):
	# 处理每个视频，返回路径和帧数
	frame_count = get_frame_count(video_path)
	# if fps < 29 or fps > 31:
	global data_type
	label_path = None
	if data_type == "hallo3":
	label_path = video_path.replace("/videos/", "/face_parse_labels/").replace(".mp4", ".mp4.mkv")
	elif data_type == "MEAD":
	label_path = video_path.replace("/MEAD/", "/MEAD_face_labels/").replace(".mp4", ".mkv")
	elif data_type == "TED":
	label_path = video_path.replace("/test_clips_tian_scenecut_resampled/", "/test_clips_tian_scenecut_face_labels_resampled_face_alignment/").replace(".mp4", ".mkv")

	if label_path and os.path.exists(label_path):
	labels = _read_labels_from_video(label_path)
	if labels is None:
	return None
	bboxes = _compute_lip_bboxes(labels)
	if _has_area_jump(bboxes, area_ratio_thresh=1.5):
	return None
	return {"video_path": video_path, "frame_count": frame_count, "label_path": label_path}

	return None
	# return {"video_path": video_path, "frame_count": frame_count, "label_path": video_path.replace("/test_clips_tian_scenecut_resampled/", "/test_clips_tian_scenecut_face_labels_resampled/").replace(".mp4", ".mkv")}

	def save_to_jsonl(results, output_file):
	# 保存处理结果到JSONL文件
	with open(output_file, 'w', encoding='utf-8') as f:
	for result in results:
	f.write(json.dumps(result, ensure_ascii=False) + '\n')

	def print_frame_distribution(frame_counts):
	# 使用Counter统计帧数分布
	frame_count_distribution = Counter(frame_counts)
	# fps_distribution = Counter(fps_counts)

	print("帧数分布情况：")
	for frame_count, count in sorted(frame_count_distribution.items()):
	print(f"帧数: {frame_count} - 文件数: {count}")

	# print("帧数分布情况：")
	# for fps_count, count in sorted(fps_distribution.items()):
	# print(f"帧数: {fps_count} - 文件数: {count}")

	def format_count_suffix(count):
	if count >= 1000:
	suffix = f"{count / 1000:.1f}k"
	return suffix.replace(".0k", "k")
	return str(count)

	def main(directory, output_file):
	# 步骤1：获取所有mp4文件路径
	video_paths = find_mp4_files(directory)

	# 步骤2：使用多进程并行处理视频文件，并添加进度条
	with multiprocessing.Pool() as pool:
	# 使用tqdm结合multiprocessing.Pool.map，显示进度条
	results = list(tqdm(pool.imap(process_video, video_paths), total=len(video_paths), desc="处理视频"))

	filter_results = [res for res in results if res is not None and res['frame_count'] > 90]

	# 步骤3：提取所有的帧数并打印帧数分布情况
	frame_counts = [result['frame_count'] for result in filter_results]
	# fps_counts = [round(result['fps']) for result in filter_results]
	print_frame_distribution(frame_counts)

	# 步骤4：保存结果到JSONL文件
	count_suffix = format_count_suffix(len(filter_results))
	base, ext = os.path.splitext(output_file)
	output_file_with_count = f"{base}_{count_suffix}{ext}"
	save_to_jsonl(filter_results, output_file_with_count)

	print(f"len results is {len(filter_results)}")
	print(f"output file is {output_file_with_count}")

	print("完成！所有的.mp4文件路径和帧数已经保存为 JSONL 格式。")

	# # 示例调用
	# directory = '/share/zhaohu_workspace/Downloaded_Data/MEAD' # 替换为目标目录路径
	# # output_file = 'MEAD_f90_facealignment.jsonl' # 输出文件名
	# output_file = 'MEAD_f90.jsonl' # 输出文件名
	# data_type = "MEAD"


	# 示例调用
	directory = '/share/zhaohu_workspace/Downloaded_Data/hallo3_training_data/videos' # 替换为目标目录路径
	output_file = 'hallo3_f90.jsonl' # 输出文件名
	data_type = "hallo3"


	# directory = '/share/zhaohu_workspace/Downloaded_Data/ted_clips/test_clips_tian_scenecut_resampled' # 替换为目标目录路径
	# data_type = "TED"
	# output_file = 'TED_f90_facealignment.jsonl' # 输出文件名

	main(directory, output_file)