📘 Overview

VideoOCREvaluator is a video OCR evaluation operator that uses PaddleOCR to detect and recognize text in video clips. It reads upstream extracted video frames, calculates the ratio of text area to frame area for each clip, and writes the OCR scores back into the video_clips field.

---

🏗️ __init__ Function

def __init__(
    self,
    figure_root: str = "extract_frames",
    input_video_key: str = "video",
    video_clips_key: str = "video_clips",
    load_num: int = 3,
    batch_size: int = 8,
    num_workers: int = 4,
    gpu_num: int = 0,
    init_distributed: bool = False,
    output_key: str = "video_clips",
    det_model_dir: str = None,
    rec_model_dir: str = None
):
    ...

🧾 __init__ Parameters

Parameter	Type	Default	Description
figure_root	str	"extract_frames"	Root directory for extracted frames
input_video_key	str	"video"	Field name for video path
video_clips_key	str	"video_clips"	Field name for video clips
load_num	int	3	Number of frames per clip to load
batch_size	int	8	Batch size for processing
num_workers	int	4	Number of data loading workers
gpu_num	int	0	GPU ID (0+ for GPU, -1 for CPU)
init_distributed	bool	False	Whether to initialize distributed training
output_key	str	"video_clips"	Output field name (updated clips)
det_model_dir	str	None	PaddleOCR detection model directory path (optional)
rec_model_dir	str	None	PaddleOCR recognition model directory path (optional)

---

⚡ run Function

def run(
    self,
    storage: DataFlowStorage,
    figure_root: Optional[str] = None,
    input_video_key: Optional[str] = None,
    video_clips_key: Optional[str] = None,
    load_num: Optional[int] = None,
    batch_size: Optional[int] = None,
    num_workers: Optional[int] = None,
    gpu_num: Optional[int] = None,
    init_distributed: Optional[bool] = None,
    output_key: Optional[str] = None,
    det_model_dir: Optional[str] = None,
    rec_model_dir: Optional[str] = None
):
    ...

Executes the main logic: reads dataframe and extracted video frames from storage, uses PaddleOCR to detect text regions, calculates text area ratio, and writes back to storage.

🧾 run Parameters

All parameters are optional and override initialization parameters. Parameter descriptions are the same as in __init__.

---

🧠 Example Usage

from dataflow.utils.storage import FileStorage
from dataflow.operators.core_vision import VideoOCREvaluator

# Step 1: Prepare FileStorage (needs video, video_clips columns)
# Note: VideoFrameFilter must be run first to extract frames
storage = FileStorage(
    first_entry_file_name="data/video_ocr_input.jsonl",
    cache_path="./cache_local",
    file_name_prefix="video_ocr",
    cache_type="jsonl"
)

# Step 2: Initialize operator
evaluator = VideoOCREvaluator(
    figure_root="./cache/extract_frames",
    input_video_key="video",
    video_clips_key="video_clips",
    load_num=3,
    batch_size=8,
    num_workers=4,
    gpu_num=0,  # Use GPU 0
    init_distributed=False
)

# Step 3: Execute evaluation
evaluator.run(
    storage=storage.step()
)

---

🧾 Default Output Format

Modified Field:

• video_clips (dict): Updated video clips dictionary, adds ocr_score to each clip

New Field in Each Clip:

Field	Type	Description
ocr_score	float	OCR score (text area ratio, 0-1, higher means more text)

Example Input:

{
  "video": ["./test/video1.mp4"],
  "video_clips": {
    "clips": [
      {
        "id": "video1_0",
        "frame_start": 0,
        "frame_end": 150,
        "num_frames": 150,
        "height": 720,
        "width": 1280
      }
    ]
  }
}

Example Output:

{
  "video": ["./test/video1.mp4"],
  "video_clips": {
    "clips": [
      {
        "id": "video1_0",
        "frame_start": 0,
        "frame_end": 150,
        "num_frames": 150,
        "height": 720,
        "width": 1280,
        "ocr_score": 0.15
      }
    ]
  }
}

---

🔗 Related Links

• Code: VideoOCREvaluator
• PaddleOCR: https://github.com/PaddlePaddle/PaddleOCR