Upload hf_job_nsfw.py with huggingface_hub

hf_job_nsfw.py

@@ -1,7 +1,8 @@
 #!/usr/bin/env python3
 """
-NSFW Classification Job - Process images with EraX YOLO
-Outputs: Per-image NSFW detections with bboxes and confidence scores
+NSFW Classification Job - Process human crops from SAM3D bboxes with EraX YOLO
+Requires: SAM 3D Body outputs for human bboxes
+Outputs: Per-human NSFW detections with bboxes and confidence scores
 """
 import argparse
 import os
@@ -22,34 +23,91 @@ logger = logging.getLogger(__name__)
 
 import numpy as np
 import torch
-from datasets import load_dataset, Dataset as HFDataset, Features, Value, Sequence
+from datasets import load_dataset, Dataset as HFDataset, Features, Value
 from PIL import Image
 import json
 from huggingface_hub import snapshot_download
 from ultralytics import YOLO
 
 
-def process_batch(batch):
-    """Process batch of images with NSFW detection"""
+def make_square_bbox_with_padding(bbox, img_width, img_height, padding=0.1):
+    """Convert bbox to square with padding"""
+    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, cx - size / 2)
+    y1_sq = max(0, cy - size / 2)
+    x2_sq = min(img_width, cx + size / 2)
+    y2_sq = min(img_height, cy + size / 2)
+    
+    return [x1_sq, y1_sq, x2_sq, y2_sq]
+
+
+def process_batch(batch, sam3d_dataset):
+    """Process batch of images - join with SAM3D results to get human bboxes"""
     images = batch['image']
     image_paths = batch.get('image_path', [f'img_{i:06d}' for i in range(len(images))])
     
     results_list = []
     
-    # Convert all images to numpy arrays
+    # Collect all crops for batch inference
     crops = []
-    for image_pil in images:
+    crop_info = []  # (image_idx, human_idx, original_bbox)
+    
+    for idx, image_pil in enumerate(images):
+        image_id = Path(image_paths[idx]).stem if image_paths[idx] else f'img_{idx:06d}'
+        
+        # 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,
+                'human_detections': None
+            })
+            continue
+        
+        humans_data = json.loads(sam3d_row[0]['sam3d_data'])
         image_rgb = np.array(image_pil.convert('RGB'))
-        crops.append(image_rgb)
+        img_width, img_height = image_pil.size
+        
+        # Collect crops for each human
+        for human_idx, human in enumerate(humans_data):
+            bbox = human.get('bbox')
+            if bbox is None:
+                continue
+            
+            # Make square bbox with padding
+            square_bbox = make_square_bbox_with_padding(bbox, img_width, img_height, padding=0.15)
+            x1, y1, x2, y2 = map(int, square_bbox)
+            
+            # Crop and resize to standard size for YOLO
+            crop = image_rgb[y1:y2, x1:x2]
+            if crop.size > 0:
+                crops.append(crop)
+                crop_info.append((idx, human_idx, square_bbox, bbox))
+    
+    # Batch NSFW inference on all crops
+    human_results = {}  # {image_idx: {human_idx: detections}}
     
-    # Batch inference
     if crops:
         try:
             yolo_results = model(crops, conf=0.2, iou=0.3, verbose=False)
             
-            for idx, result in enumerate(yolo_results):
-                image_id = Path(image_paths[idx]).stem if image_paths[idx] else f'img_{idx:06d}'
-                img_width, img_height = images[idx].size
+            for crop_idx, result in enumerate(yolo_results):
+                img_idx, human_idx, square_bbox, orig_bbox = crop_info[crop_idx]
+                x1_sq, y1_sq, x2_sq, y2_sq = square_bbox
                 
                 detections = []
                 if result.boxes:
@@ -59,38 +117,53 @@ def process_batch(batch):
                         class_names = ['anus', 'make_love', 'nipple', 'penis', 'vagina']
                         class_name = class_names[class_id] if class_id < len(class_names) else f'class_{class_id}'
                         
-                        x1, y1, x2, y2 = box.xyxy[0].tolist()
+                        # Convert crop coordinates to image coordinates
+                        dx1, dy1, dx2, dy2 = box.xyxy[0].tolist()
+                        abs_x1 = x1_sq + dx1
+                        abs_y1 = y1_sq + dy1
+                        abs_x2 = x1_sq + dx2
+                        abs_y2 = y1_sq + dy2
                         
                         detections.append({
                             'class': class_name,
                             'confidence': confidence,
-                            'bbox': [x1, y1, x2, y2]
+                            'bbox': [abs_x1, abs_y1, abs_x2, abs_y2]
                         })
                 
                 if not detections:
-                    detections = [{'class': 'safe', 'confidence': 1.0, 'bbox': [0, 0, img_width, img_height]}]
+                    detections = [{'class': 'safe', 'confidence': 1.0, 'bbox': orig_bbox}]
                 
-                results_list.append({
-                    'image_id': image_id,
-                    'detections': detections,
-                    'num_detections': len(detections)
-                })
+                if img_idx not in human_results:
+                    human_results[img_idx] = {}
+                human_results[img_idx][human_idx] = detections
+        
         except Exception as e:
             logger.error(f"NSFW batch failed: {e}")
-            # Fallback: mark all as safe
-            for idx, image_path in enumerate(image_paths):
-                image_id = Path(image_path).stem if image_path else f'img_{idx:06d}'
-                img_width, img_height = images[idx].size
-                results_list.append({
-                    'image_id': image_id,
-                    'detections': [{'class': 'safe', 'confidence': 1.0, 'bbox': [0, 0, img_width, img_height]}],
-                    'num_detections': 1
-                })
+    
+    # Organize results by image
+    for idx, image_path in enumerate(image_paths):
+        image_id = Path(image_path).stem if image_path else f'img_{idx:06d}'
+        
+        if idx in human_results:
+            # Convert dict to list ordered by human_idx
+            max_human_idx = max(human_results[idx].keys())
+            detections_list = []
+            for h_idx in range(max_human_idx + 1):
+                detections_list.append(human_results[idx].get(h_idx, [{'class': 'safe', 'confidence': 1.0}]))
+            
+            results_list.append({
+                'image_id': image_id,
+                'human_detections': json.dumps(detections_list)
+            })
+        else:
+            results_list.append({
+                'image_id': image_id,
+                'human_detections': None
+            })
     
     return {
         'image_id': [r['image_id'] for r in results_list],
-        'detections_json': [json.dumps(r['detections']) for r in results_list],
-        'num_detections': [r['num_detections'] for r in results_list]
+        'nsfw_detections': [r['human_detections'] for r in results_list]
     }
 
 
@@ -98,14 +171,15 @@ def main():
     global model
     
     logger.info("="*60)
-    logger.info("NSFW Classification with EraX YOLO")
+    logger.info("NSFW Classification with EraX YOLO (Per-Human)")
     logger.info("="*60)
     
     ap = argparse.ArgumentParser()
-    ap.add_argument('--input-dataset', type=str, required=True)
+    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=8)
+    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()
@@ -123,19 +197,27 @@ def main():
     model = YOLO('erax_nsfw_yolo11m.pt')
     logger.info("✓ Model loaded")
     
-    # Load dataset
-    logger.info(f"Loading dataset {args.input_dataset}...")
+    # 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 with batching
     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_batch,
+        process_fn,
         batched=True,
         batch_size=args.batch_size,
         remove_columns=ds.column_names
@@ -154,14 +236,12 @@ def main():
     # Create output dataset
     features = Features({
         'image_id': Value('string'),
-        'detections_json': Value('string'),
-        'num_detections': Value('int32')
+        'nsfw_detections': Value('string')
     })
     
     output_ds = HFDataset.from_dict({
         'image_id': [r['image_id'] for r in results],
-        'detections_json': [r['detections_json'] for r in results],
-        'num_detections': [r['num_detections'] for r in results]
+        'nsfw_detections': [r['nsfw_detections'] for r in results]
     }, features=features)
     
     # Upload