File size: 9,853 Bytes

#!/usr/bin/env python3
"""
Face Embedding Job - Extract ArcFace embeddings from detected faces
Requires: SAM 3D Body outputs for face bboxes
Outputs: 512-dim face embeddings with detection confidence
"""
import argparse
import os
from pathlib import Path
import warnings
warnings.filterwarnings('ignore')
import logging
import sys
import subprocess

logging.basicConfig(
    level=logging.INFO,
    format='[%(asctime)s] %(levelname)s: %(message)s',
    datefmt='%Y-%m-%d %H:%M:%S',
    stream=sys.stdout,
    force=True
)
logger = logging.getLogger(__name__)

import numpy as np
import torch
from datasets import load_dataset, Dataset as HFDataset, Features, Value
from PIL import Image
import cv2
import json


def init_face_embedder(device='cuda'):
    """Initialize InsightFace ArcFace model"""
    logger.info("Installing InsightFace...")
    try:
        subprocess.run(
            ['pip', 'install', '-q', 'insightface', 'onnxruntime-gpu' if device.type == 'cuda' else 'onnxruntime'],
            check=True,
            capture_output=True
        )
        logger.info("✓ InsightFace installed")
    except Exception as e:
        logger.warning(f"InsightFace installation failed: {e}")
    
    logger.info("Loading InsightFace ArcFace...")
    import insightface
    from insightface.app import FaceAnalysis
    
    app = FaceAnalysis(
        name='buffalo_l',
        providers=['CUDAExecutionProvider'] if device.type == 'cuda' else ['CPUExecutionProvider']
    )
    app.prepare(ctx_id=0 if device.type == 'cuda' else -1, det_size=(640, 640))
    logger.info("✓ ArcFace loaded")
    
    return app


def make_square_bbox_with_padding(bbox, img_width, img_height, padding=0.2):
    """Convert bbox to square with padding for face detection"""
    x1, y1, x2, y2 = bbox
    w = x2 - x1
    h = y2 - y1
    
    # Make square
    size = max(w, h)
    cx = (x1 + x2) / 2
    cy = (y1 + y2) / 2
    
    # Add padding
    size = size * (1 + padding)
    
    # Get square bbox
    x1_sq = max(0, int(cx - size / 2))
    y1_sq = max(0, int(cy - size / 2))
    x2_sq = min(img_width, int(cx + size / 2))
    y2_sq = min(img_height, int(cy + size / 2))
    
    return [x1_sq, y1_sq, x2_sq, y2_sq]


def has_valid_face(keypoints_2d, keypoints_3d, img_width, img_height):
    """Check if human has a valid, visible face"""
    if keypoints_2d is None or keypoints_3d is None:
        return False
    
    kpts2d_arr = np.array(keypoints_2d)
    kpts3d_arr = np.array(keypoints_3d)
    
    if len(kpts2d_arr) < 3 or len(kpts3d_arr) < 3:
        return False
    
    # Check face keypoints (nose, left eye, right eye)
    nose_2d = kpts2d_arr[0]
    left_eye_2d = kpts2d_arr[1]
    right_eye_2d = kpts2d_arr[2]
    nose_3d = kpts3d_arr[0]
    left_eye_3d = kpts3d_arr[1]
    right_eye_3d = kpts3d_arr[2]
    
    # Check 3D keypoints are valid (not at origin)
    keypoints_valid_3d = (np.linalg.norm(nose_3d) > 1e-6 and 
                         np.linalg.norm(left_eye_3d) > 1e-6 and 
                         np.linalg.norm(right_eye_3d) > 1e-6)
    
    if not keypoints_valid_3d:
        return False
    
    # Check 2D keypoints are within image bounds
    for kp in [nose_2d, left_eye_2d, right_eye_2d]:
        if (kp[0] < 0 or kp[0] >= img_width or 
            kp[1] < 0 or kp[1] >= img_height):
            return False
    
    return True


def extract_embedding(app, image_bgr, bbox, img_width, img_height):
    """Extract face embedding from bbox region with proper cropping and padding"""
    try:
        # Make square bbox with padding for better face detection
        square_bbox = make_square_bbox_with_padding(bbox, img_width, img_height, padding=0.2)
        x1, y1, x2, y2 = square_bbox
        
        # Crop to square region
        crop = image_bgr[y1:y2, x1:x2]
        
        if crop.size == 0:
            return None
        
        # Resize to optimal size for InsightFace (640x640 max)
        crop_h, crop_w = crop.shape[:2]
        if max(crop_h, crop_w) > 640:
            scale = 640 / max(crop_h, crop_w)
            new_h = int(crop_h * scale)
            new_w = int(crop_w * scale)
            crop = cv2.resize(crop, (new_w, new_h), interpolation=cv2.INTER_LINEAR)
        
        # Detect faces
        faces = app.get(crop)
        if len(faces) == 0:
            return None
        
        # Use the most confident face
        face = max(faces, key=lambda x: x.det_score)
        embedding = face.embedding
        embedding_norm = embedding / np.linalg.norm(embedding)
        
        return {
            'embedding': embedding_norm.astype(np.float32).tolist(),
            'det_score': float(face.det_score),
            'embedding_dim': len(embedding)
        }
    except Exception as e:
        logger.error(f"Embedding extraction failed: {e}")
        return None


def process_batch(batch, sam3d_dataset):
    """Process batch of images - join with SAM3D results to get bboxes"""
    images = batch['image']
    image_paths = batch.get('image_path', [f'img_{i:06d}' for i in range(len(images))])
    
    results_list = []
    
    for idx, image_pil in enumerate(images):
        image_id = Path(image_paths[idx]).stem if image_paths[idx] else f'img_{idx:06d}'
        img_width, img_height = image_pil.size
        
        # Find corresponding SAM3D data
        sam3d_row = sam3d_dataset.filter(lambda x: x['image_id'] == image_id).take(1)
        sam3d_row = list(sam3d_row)
        
        if not sam3d_row or not sam3d_row[0]['sam3d_data']:
            results_list.append({
                'image_id': image_id,
                'embeddings': None
            })
            continue
        
        humans_data = json.loads(sam3d_row[0]['sam3d_data'])
        
        # Convert to BGR
        image_rgb = np.array(image_pil.convert('RGB'))
        image_bgr = cv2.cvtColor(image_rgb, cv2.COLOR_RGB2BGR)
        
        # Extract embeddings for each human with valid face
        embeddings = []
        for human_idx, human in enumerate(humans_data):
            bbox = human.get('bbox')
            kpts2d = human.get('keypoints_2d')
            kpts3d = human.get('keypoints_3d')
            
            # Check if this human has a valid, visible face
            if not has_valid_face(kpts2d, kpts3d, img_width, img_height):
                embeddings.append(None)
                continue
            
            if bbox is None:
                embeddings.append(None)
                continue
            
            # Extract embedding from face region
            embedding = extract_embedding(face_app, image_bgr, bbox, img_width, img_height)
            embeddings.append(embedding)
        
        results_list.append({
            'image_id': image_id,
            'embeddings': json.dumps(embeddings) if any(e is not None for e in embeddings) else None
        })
    
    return {
        'image_id': [r['image_id'] for r in results_list],
        'face_embeddings': [r['embeddings'] for r in results_list]
    }


def main():
    global face_app
    
    logger.info("="*60)
    logger.info("Face Embedding Extraction (ArcFace)")
    logger.info("="*60)
    
    ap = argparse.ArgumentParser()
    ap.add_argument('--input-dataset', type=str, required=True, help='Original images')
    ap.add_argument('--sam3d-dataset', type=str, required=True, help='SAM3D outputs with bboxes')
    ap.add_argument('--output-dataset', type=str, required=True)
    ap.add_argument('--split', type=str, default='train')
    ap.add_argument('--batch-size', type=int, default=4)
    ap.add_argument('--shard-index', type=int, default=0)
    ap.add_argument('--num-shards', type=int, default=1)
    args = ap.parse_args()
    
    logger.info(f"Arguments: {vars(args)}")
    
    device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
    logger.info(f"Using device: {device}")
    
    # Load face embedder
    face_app = init_face_embedder(device)
    
    # Load SAM3D results
    logger.info(f"Loading SAM3D results from {args.sam3d_dataset}...")
    sam3d_ds = load_dataset(args.sam3d_dataset, split=args.split, streaming=True)
    
    # Load images dataset
    logger.info(f"Loading images from {args.input_dataset}...")
    ds = load_dataset(args.input_dataset, split=args.split, streaming=True)
    
    if args.num_shards > 1:
        ds = ds.shard(num_shards=args.num_shards, index=args.shard_index)
        sam3d_ds = sam3d_ds.shard(num_shards=args.num_shards, index=args.shard_index)
        logger.info(f"Using shard {args.shard_index+1}/{args.num_shards}")
    
    # Process
    logger.info(f"Processing with batch_size={args.batch_size}")
    
    from functools import partial
    process_fn = partial(process_batch, sam3d_dataset=sam3d_ds)
    
    processed_ds = ds.map(
        process_fn,
        batched=True,
        batch_size=args.batch_size,
        remove_columns=ds.column_names
    )
    
    # Collect results
    results = []
    for batch_idx, item in enumerate(processed_ds):
        results.append(item)
        
        if (batch_idx + 1) % 50 == 0:
            logger.info(f"Processed {batch_idx + 1} images")
    
    logger.info(f"✓ Processed {len(results)} images")
    
    # Create output dataset
    features = Features({
        'image_id': Value('string'),
        'face_embeddings': Value('string')
    })
    
    output_ds = HFDataset.from_dict({
        'image_id': [r['image_id'] for r in results],
        'face_embeddings': [r['face_embeddings'] for r in results]
    }, features=features)
    
    # Upload
    logger.info(f"Uploading to {args.output_dataset}...")
    output_ds.push_to_hub(
        args.output_dataset,
        split=args.split,
        token=os.environ.get('HF_TOKEN'),
        private=True
    )
    logger.info("✓ Upload complete")


if __name__ == '__main__':
    main()