File size: 19,326 Bytes
3eacd49 b5958b1 8e74163 b5958b1 3e0a7cf 23ba375 8e74163 b5958b1 8e74163 b5958b1 8e74163 b5958b1 8e74163 60fde3b 23ba375 b5958b1 8e74163 60fde3b 8e74163 60fde3b 8e74163 60fde3b 8e74163 60fde3b 8e74163 60fde3b b5958b1 8e74163 60fde3b 8e74163 23ba375 8e74163 23ba375 8e74163 23ba375 8e74163 23ba375 8e74163 23ba375 8e74163 23ba375 8e74163 23ba375 8e74163 60fde3b 8e74163 60fde3b 8e74163 23ba375 8e74163 23ba375 8e74163 23ba375 8e74163 23ba375 8e74163 23ba375 60fde3b 23ba375 60fde3b 23ba375 60fde3b 23ba375 60fde3b 23ba375 60fde3b 23ba375 60fde3b 23ba375 60fde3b 23ba375 60fde3b 23ba375 60fde3b 23ba375 60fde3b 23ba375 60fde3b 23ba375 60fde3b 23ba375 60fde3b 23ba375 8e74163 23ba375 b5958b1 8e74163 b5958b1 8e74163 23ba375 8e74163 3eacd49 8e74163 60fde3b 8e74163 60fde3b 8e74163 3eacd49 23ba375 8e74163 b5958b1 8e74163 3e0a7cf b5958b1 8e74163 b5958b1 8e74163 b5958b1 8e74163 b5958b1 8e74163 b5958b1 8e74163 b5958b1 8e74163 b5958b1 8e74163 3e0a7cf 8e74163 3eacd49 b5958b1 | 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 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 | """
**任务名称:** 围绕静止物体的相机视角排序(Camera View Ordering around Static Objects)
**任务设定:**
本任务模拟一个静止物体被移动相机拍摄的场景。相机围绕物体进行平滑的轨道旋转(Orbital Rotation)。
1. **数据来源:** 所有图像均从一段连续的视频中抽帧获得,因此具备内在的时空连贯性。
2. **运动约束:** 从给定的初始参考帧(Reference Frame)到序列的最后一帧,相机的总旋转角度严格限制在半个圆周(<180度)以内,保证视角的单向性和连续性。
**输入数据:**
* 一张**初始参考图(Reference Image)**,作为序列的起点(t=0)。
* 四张**候选图片(Candidate Images)**,编号为 1、2、3、4。这四张图片是参考图之后的后续帧,但顺序已被打乱。
**任务目标:**
模型需要分析参考图与候选图片之间的几何关系,将四张候选图片按照真实的时间/空间顺序重新排列,使其构成一个连贯的相机运动轨迹。
**输出格式:**
提供 A、B、C、D 四个选项,每个选项代表一种排序组合(例如 B: 2-1-4-3)。模型需选择能够恢复正确时空顺序的选项。
"""
import argparse
import random
import itertools
import json
import os
from tqdm import tqdm
import numpy as np
# 导入公共工具库
from utils import (
CO3DDataLoader,
get_relative_yaw,
format_image_path,
save_jsonl_splits,
get_sequence_geometry_pca,
decompose_angle,
)
class Task3Generator:
def __init__(self, loader, image_prefix):
self.loader = loader
self.image_prefix = image_prefix
self.cat_name = self.loader.category.replace('_', ' ')
def _get_all_relative_angles(self, start_idx, all_frames, seq_data_dict, mean_center, basis):
"""
计算序列中所有帧相对于 start_idx 的角度。
Args:
start_idx: 起始帧索引
all_frames: 所有帧索引列表
seq_data_dict: 对齐后的序列数据 {frame_idx: {'R': ..., 'T': ..., 'path': ...}}
mean_center: 场景中心
basis: 基底矩阵
Returns:
列表: [{'idx': frame_idx, 'angle': relative_angle}, ...]
"""
start_info = seq_data_dict[start_idx]
results = []
for f_idx in all_frames:
if f_idx == start_idx:
results.append({'idx': f_idx, 'angle': 0.0})
continue
f_info = seq_data_dict[f_idx]
yaw = get_relative_yaw(
start_info['R'], start_info['T'],
f_info['R'], f_info['T'],
mean_center, basis
)
results.append({'idx': f_idx, 'angle': yaw})
return results
def generate_sample(self, seq_name, config):
# 获取该序列所有帧并排序
frames = sorted(self.loader.get_frames(seq_name))
if len(frames) < 10: # 稍微提高一点门槛,太短的序列很难凑齐5张符合间隔的图
return None
# 获取几何信息(现在返回三个值)
seq_data_dict = self.loader.seq_data[seq_name]
mean_center, basis, _ = get_sequence_geometry_pca(seq_data_dict)
min_int = config['min_interval']
max_int = config['max_interval']
max_span = config['max_angle']
max_attempts = 5000 # 尝试寻找起点的次数
for _ in range(max_attempts):
# 1. 随机选择一个起点 (Reference Frame)
# 留出后面至少4帧的空间,虽然不一定连续,但索引得靠前
start_idx = random.choice(frames[:-4])
# 2. 计算所有后续帧相对于起点的角度(使用对齐后的数据)
rel_data = self._get_all_relative_angles(
start_idx, frames, seq_data_dict, mean_center, basis
)
# 过滤掉 start 之前的帧
future_candidates = [x for x in rel_data if x['idx'] > start_idx]
# 3. 逐步构建链条 (Chain Building)
# 我们需要找到 [Ref, C1, C2, C3, C4]
# 使得 abs(Angle(Ci) - Angle(Ci-1)) 在 [min_int, max_int] 之间
chain = [{'idx': start_idx, 'angle': 0.0}]
for step in range(4): # 需要找 4 个后续帧
last_item = chain[-1]
last_angle = last_item['angle']
valid_next_frames = []
for cand in future_candidates:
# 必须保持时间顺序 (index 递增)
if cand['idx'] <= last_item['idx']:
continue
# 计算与上一帧的间隔
diff = abs(cand['angle'] - last_angle)
# 核心约束检查:
# 1. 间隔在 [min, max] 之间
# 2. 总跨度不超过 max_span (相对于 Ref 的 0 度)
# 3. 单调性检查: 新的角度绝对值 必须 大于 旧的角度绝对值 (确保是越转越远,而不是转回来了)
if (min_int <= diff <= max_int) and \
(abs(cand['angle']) <= max_span) and \
(abs(cand['angle']) > abs(last_angle)):
valid_next_frames.append(cand)
if not valid_next_frames:
break # 链条断了,当前 start_idx 无法构成序列
# 随机选一个符合条件的下一帧
next_frame = random.choice(valid_next_frames)
chain.append(next_frame)
# 4. 检查是否成功找到了 5 帧 (Ref + 4 Candidates)
if len(chain) == 5:
# 提取索引和角度
indices = [x['idx'] for x in chain]
angles = [x['angle'] for x in chain]
return self.create_entry(seq_name, indices, angles, rel_data)
return None
def create_entry(self, seq_name, frame_indices, angles, all_rel_data):
"""
Args:
frame_indices: [Ref, True_1, True_2, True_3, True_4]
angles: [0, Ang1, Ang2, Ang3, Ang4] (相对于 Ref)
all_rel_data: [{'idx':..., 'angle':...}, ...] 序列中所有帧的角度信息
"""
# ================= 配置区域 =================
ROTATION_STEPS = [15, 10] # 优先使用 15度,补齐用 10度
# ===========================================
ref_idx = frame_indices[0]
true_candidates_indices = frame_indices[1:]
ref_info = self.loader.get_frame_info(seq_name, ref_idx)
candidates_infos = [self.loader.get_frame_info(seq_name, idx) for idx in true_candidates_indices]
# 1. 准备候选图片数据 (用于最终选项)
labeled_candidates = []
for i, info in enumerate(candidates_infos):
labeled_candidates.append({
"true_order": i + 1,
"path": format_image_path(info['path'], self.loader.root_path, self.image_prefix),
"angle_rel_ref": angles[i+1]
})
# 打乱顺序
shuffled_candidates = labeled_candidates.copy()
random.shuffle(shuffled_candidates)
# 初始化 images 字典
images_dict = {
"image_ref": format_image_path(ref_info['path'], self.loader.root_path, self.image_prefix)
}
display_mapping = {}
candidates_meta = {}
for k, item in enumerate(shuffled_candidates):
display_label = k + 1
images_dict[f"image_{display_label}"] = item['path']
display_mapping[display_label] = item['true_order']
# <--- [新增] 记录该 Display Image 的详细元数据
candidates_meta[str(display_label)] = {
"angle_relative_to_ref": item['angle_rel_ref'],
"true_rank": item['true_order'],
"image_key": f"image_{display_label}"
}
# 2. 生成选项文本 (A/B/C/D)
true_to_display = {v: k for k, v in display_mapping.items()}
correct_sequence = [true_to_display[i] for i in range(1, 5)]
correct_str = "-".join(map(str, correct_sequence))
all_permutations = list(itertools.permutations([1, 2, 3, 4]))
distractor_pool = [
"-".join(map(str, p))
for p in all_permutations
if list(p) != correct_sequence
]
selected_distractors = random.sample(distractor_pool, 3)
options_data = [{"seq": correct_str, "is_correct": True}]
for d in selected_distractors:
options_data.append({"seq": d, "is_correct": False})
random.shuffle(options_data)
option_labels = ['A', 'B', 'C', 'D']
gt_option_label = ""
sequence_options_text = {}
for label, opt in zip(option_labels, options_data):
sequence_options_text[f"sequence_{label}"] = opt["seq"]
if opt["is_correct"]:
gt_option_label = label
# ------------------------------------------------------------------
# 3. 构建 Oracle Meta 和 Chain (替代原本的文本生成)
# ------------------------------------------------------------------
is_clockwise = angles[-1] < 0
direction_str = "clockwise" if is_clockwise else "counter-clockwise"
chain = []
images_dict_update = {} # 用于存储推理过程中的中间图
current_simulated_angle = 0.0
prev_target_angle = 0.0
reasoning_img_counter = 0
# 遍历 4 个目标阶段 (Ref -> C1, C1 -> C2, ...)
# frame_indices[0] 是 Ref, frame_indices[1:] 是 C1, C2, C3, C4
for i in range(1, 5):
target_angle = angles[i]
true_delta = abs(target_angle - prev_target_angle)
# 拆解步长为 [30, 15]
steps = decompose_angle(true_delta, [30, 15])
matched_display_label = true_to_display[i] # 获取当前真实第i张图对应用户看到的 Label (1-4)
for step_idx, step_deg in enumerate(steps):
reasoning_img_counter += 1
# 1. 更新模拟角度
if is_clockwise:
current_simulated_angle -= step_deg
else:
current_simulated_angle += step_deg
# 2. 寻找最接近该角度的真实帧
closest_frame_data = min(all_rel_data, key=lambda x: abs(x['angle'] - current_simulated_angle))
closest_frame_idx = closest_frame_data['idx']
closest_frame_info = self.loader.get_frame_info(seq_name, closest_frame_idx)
# 3. 【关键修改】始终生成 reasoning_image
# 不管是不是最后一步,都把当前视角的图存下来
result_key = f"reasoning_image_{reasoning_img_counter}"
images_dict_update[result_key] = format_image_path(
closest_frame_info['path'], self.loader.root_path, self.image_prefix
)
# 4. 判断是否到达匹配点
is_last_step_of_phase = (step_idx == len(steps) - 1)
# 4. 构建 Chain Item
chain_item = {
"step_index": len(chain) + 1,
"action": {
"type": "rotate",
"degrees": step_deg,
"direction": direction_str,
"total_angle_so_far": current_simulated_angle
},
"result_image_key": result_key,
"is_key_frame": is_last_step_of_phase,
"matched_display_label": matched_display_label if is_last_step_of_phase else None
}
chain.append(chain_item)
prev_target_angle = target_angle
# 合并图片字典
images_dict.update(images_dict_update)
# ------------------------------------------------------------------
# 4. 生成 Question Prompt (保持不变)
# ------------------------------------------------------------------
template_id = random.choice([1, 2])
obj_name = self.cat_name
if template_id == 1:
question = f"""The {obj_name} in the initial view <image_start>[image_ref]<image_end> remains **static**. Imagine a camera rotating around this {obj_name} along a continuous path. The direction of rotation is defined from a **top-down bird's-eye view**. The total rotation from the start to the end is less than 180 degrees.
Below are four images captured during this rotation, labeled 1, 2, 3 and 4. They're currently shuffled.
1. <image_start>[image_1]<image_end>
2. <image_start>[image_2]<image_end>
3. <image_start>[image_3]<image_end>
4. <image_start>[image_4]<image_end>
Please analyze the change in perspective and determine the correct chronological order of these four images following the initial view. Select the correct sequence.
A. {sequence_options_text['sequence_A']}
B. {sequence_options_text['sequence_B']}
C. {sequence_options_text['sequence_C']}
D. {sequence_options_text['sequence_D']}"""
else:
question = f"""Given the initial view of the **static** {obj_name}: <image_start>[image_ref]<image_end>.
Imagine a camera rotating around this {obj_name} to capture a video sequence. The rotation covers an angle of less than 180 degrees. We have extracted four frames from the sequence, labeled 1 to 4, but their order is jumbled.
1. <image_start>[image_1]<image_end>
2. <image_start>[image_2]<image_end>
3. <image_start>[image_3]<image_end>
4. <image_start>[image_4]<image_end>
Which of the following four options correctly sorts these images into a coherent spatio-temporal sequence starting after the initial view?
A. {sequence_options_text['sequence_A']}
B. {sequence_options_text['sequence_B']}
C. {sequence_options_text['sequence_C']}
D. {sequence_options_text['sequence_D']}"""
return {
"id": f"task3_{seq_name}_{ref_idx}",
"task": "camera_view_ordering",
"sequence": seq_name,
"category": self.loader.category,
"question": question,
"images": images_dict,
"oracle_meta": {
"direction": direction_str,
"correct_sequence_str": correct_str, # e.g. "2-1-4-3"
"correct_label": gt_option_label, # e.g. "B"
"chain": chain,
"candidates_meta": candidates_meta
},
"gt_answer": f"<answer>{gt_option_label}</answer>" # 占位,后面会被 Generator 覆盖
}
def main():
parser = argparse.ArgumentParser(description="Generate Task 3: Camera View Ordering")
# 路径配置
parser.add_argument("--root_path", type=str, required=True, help="CO3D dataset root")
parser.add_argument("--output_dir", type=str, default="output_task3", help="Output directory")
parser.add_argument("--image_prefix", type=str, default="data/", help="Prefix for image paths")
# [新增] 筛选名单路径
parser.add_argument("--filter_path", type=str, default=None, help="Root directory for filter logs (containing category/keep.json)")
# 采样配置
parser.add_argument("--category", type=str, default=None, help="Specific category or None for all")
parser.add_argument("--num_samples", type=int, default=1, help="Samples per sequence")
parser.add_argument("--seed", type=int, default=42)
# 几何约束配置
# min_interval: 相邻两帧之间至少间隔多少度
parser.add_argument("--min_interval", type=float, default=15.0)
# max_interval: 相邻两帧之间最大间隔多少度 (防止跨度太大)
parser.add_argument("--max_interval", type=float, default=45.0)
# max_angle: 整个序列(Ref -> Last)的最大跨度,必须 < 180
parser.add_argument("--max_angle", type=float, default=175.0)
# 切分配置
parser.add_argument("--train_ratio", type=float, default=0.8)
parser.add_argument("--val_ratio", type=float, default=0.1)
parser.add_argument("--test_ratio", type=float, default=0.1)
parser.add_argument("--max_items", type=int, default=10000)
args = parser.parse_args()
# 初始化
random.seed(args.seed)
np.random.seed(args.seed)
if args.category:
categories = [args.category]
else:
data_dir = os.path.join(args.root_path, 'data', 'original')
if os.path.exists(data_dir):
categories = sorted([d for d in os.listdir(data_dir) if os.path.isdir(os.path.join(data_dir, d))])
else:
print(f"Error: {data_dir} not found.")
return
all_results = []
config = {
'min_interval': args.min_interval,
'max_interval': args.max_interval,
'max_angle': args.max_angle
}
for cat in categories:
loader = CO3DDataLoader(args.root_path, cat)
if not loader.seq_data:
continue
generator = Task3Generator(loader, args.image_prefix)
sequences = loader.get_sequences()
# [新增] 过滤逻辑
if args.filter_path:
keep_file = os.path.join(args.filter_path, cat, "keep.json")
if os.path.exists(keep_file):
try:
with open(keep_file, 'r') as f:
keep_list = set(json.load(f)) # 使用 set 加速查找
original_count = len(sequences)
sequences = [s for s in sequences if s in keep_list]
print(f"[{cat}] Filter applied: {original_count} -> {len(sequences)} sequences retained.")
except Exception as e:
print(f"[{cat}] Error reading keep.json: {e}. Skipping category.")
sequences = []
else:
print(f"[{cat}] Warning: No keep.json found at {keep_file}. Skipping category.")
sequences = []
if not sequences:
continue
for seq in tqdm(sequences, desc=f"Task3 - {cat}", leave=False):
for _ in range(args.num_samples):
sample = generator.generate_sample(seq, config)
if sample:
all_results.append(sample)
print(f"Total generated: {len(all_results)}")
save_jsonl_splits(
all_results,
args.output_dir,
ratios=(args.train_ratio, args.val_ratio, args.test_ratio),
max_items=args.max_items,
seed=args.seed
)
print(f"Done. Output saved to {args.output_dir}")
if __name__ == "__main__":
main()
|