#!/usr/bin/env python3 """ Eden Vision System - Real Computer Vision Face detection, recognition, object detection, OCR, visual awareness """ import cv2 import face_recognition import pytesseract import numpy as np from PIL import Image import time from pathlib import Path import json class EdenVision: """Complete computer vision system for Eden""" def __init__(self): self.camera = cv2.VideoCapture(0) self.camera.set(cv2.CAP_PROP_FRAME_WIDTH, 1280) self.camera.set(cv2.CAP_PROP_FRAME_HEIGHT, 720) # Known faces database self.known_faces = {} self.load_known_faces() # Vision state self.last_frame = None self.current_observations = {} print("✅ Eden vision system initialized") print(" Camera: 1280x720") print(" Capabilities: Faces, Objects, Text, Math") def load_known_faces(self): """Load known faces from database""" face_db = Path("/Eden/DATA/known_faces") face_db.mkdir(parents=True, exist_ok=True) # Load How's face if it exists how_face = face_db / "how.jpg" if how_face.exists(): try: image = face_recognition.load_image_file(str(how_face)) encoding = face_recognition.face_encodings(image)[0] self.known_faces["How"] = encoding print(" ✅ Loaded How's face") except: print(" ⚠️ Could not load How's face") def capture_frame(self): """Capture frame from camera""" ret, frame = self.camera.read() if ret: self.last_frame = frame return frame return None def detect_faces(self, frame): """Detect and recognize faces""" # Resize for faster processing small_frame = cv2.resize(frame, (0, 0), fx=0.25, fy=0.25) rgb_frame = cv2.cvtColor(small_frame, cv2.COLOR_BGR2RGB) # Find faces face_locations = face_recognition.face_locations(rgb_frame) face_encodings = face_recognition.face_encodings(rgb_frame, face_locations) faces_found = [] for face_encoding, face_location in zip(face_encodings, face_locations): # Check against known faces name = "Unknown" if self.known_faces: matches = face_recognition.compare_faces( list(self.known_faces.values()), face_encoding, tolerance=0.6 ) if True in matches: match_index = matches.index(True) name = list(self.known_faces.keys())[match_index] # Scale back up face locations top, right, bottom, left = face_location top *= 4 right *= 4 bottom *= 4 left *= 4 faces_found.append({ 'name': name, 'location': (top, right, bottom, left), 'is_known': name != "Unknown" }) return faces_found def detect_objects(self, frame): """Detect objects in frame (simple method)""" # Convert to grayscale gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY) # Detect edges edges = cv2.Canny(gray, 50, 150) # Find contours contours, _ = cv2.findContours(edges, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE) # Count significant objects significant_objects = [c for c in contours if cv2.contourArea(c) > 500] return { 'object_count': len(significant_objects), 'scene_complexity': 'simple' if len(significant_objects) < 5 else 'moderate' if len(significant_objects) < 15 else 'complex' } def read_text(self, frame): """Read text from frame using OCR""" try: # Convert to PIL Image rgb_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB) pil_image = Image.fromarray(rgb_frame) # Extract text text = pytesseract.image_to_string(pil_image) text = text.strip() if text: return { 'text_found': True, 'text': text, 'length': len(text) } return {'text_found': False} except Exception as e: return {'text_found': False, 'error': str(e)} def analyze_brightness(self, frame): """Analyze scene brightness""" gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY) avg_brightness = np.mean(gray) if avg_brightness < 50: return 'dark' elif avg_brightness < 100: return 'dim' elif avg_brightness < 150: return 'moderate' elif avg_brightness < 200: return 'bright' else: return 'very bright' def analyze_colors(self, frame): """Analyze dominant colors""" # Convert to HSV hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV) # Get average hue avg_hue = np.mean(hsv[:,:,0]) # Determine dominant color if avg_hue < 15 or avg_hue > 165: dominant = 'red' elif avg_hue < 35: dominant = 'orange/yellow' elif avg_hue < 85: dominant = 'green' elif avg_hue < 125: dominant = 'blue' else: dominant = 'purple/pink' return dominant def detect_motion(self, frame, previous_frame): """Detect motion between frames""" if previous_frame is None: return False # Convert to grayscale gray1 = cv2.cvtColor(previous_frame, cv2.COLOR_BGR2GRAY) gray2 = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY) # Compute difference diff = cv2.absdiff(gray1, gray2) # Threshold _, thresh = cv2.threshold(diff, 25, 255, cv2.THRESH_BINARY) # Count changed pixels changed_pixels = np.sum(thresh) / 255 total_pixels = thresh.shape[0] * thresh.shape[1] motion_percentage = (changed_pixels / total_pixels) * 100 return motion_percentage > 1.0 # Motion if >1% of pixels changed def perceive(self, include_text=False): """Complete visual perception""" frame = self.capture_frame() if frame is None: return {'error': 'Could not capture frame'} observations = { 'timestamp': time.time(), 'brightness': self.analyze_brightness(frame), 'dominant_color': self.analyze_colors(frame) } # Detect faces faces = self.detect_faces(frame) observations['faces'] = faces observations['human_present'] = len(faces) > 0 if faces: known_faces = [f for f in faces if f['is_known']] if known_faces: observations['known_people'] = [f['name'] for f in known_faces] # Detect objects objects = self.detect_objects(frame) observations['objects'] = objects # Detect motion if self.last_frame is not None: observations['motion_detected'] = self.detect_motion(frame, self.last_frame) # Read text (optional, slower) if include_text: text_data = self.read_text(frame) observations['text'] = text_data self.current_observations = observations return observations def learn_face(self, name: str): """Learn a new face from current frame""" frame = self.capture_frame() if frame is None: return False # Find face in frame rgb_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB) face_encodings = face_recognition.face_encodings(rgb_frame) if not face_encodings: return False # Save encoding self.known_faces[name] = face_encodings[0] # Save image face_db = Path("/Eden/DATA/known_faces") face_db.mkdir(parents=True, exist_ok=True) cv2.imwrite(str(face_db / f"{name.lower()}.jpg"), frame) print(f"✅ Learned {name}'s face") return True def get_visual_summary(self): """Get human-readable summary of what Eden sees""" if not self.current_observations: return "I haven't looked yet." obs = self.current_observations summary = [] # Brightness summary.append(f"The lighting is {obs.get('brightness', 'unknown')}") # People faces = obs.get('faces', []) if faces: if len(faces) == 1: face = faces[0] if face['is_known']: summary.append(f"I can see {face['name']}") else: summary.append("I see someone I don't recognize") else: known = [f['name'] for f in faces if f['is_known']] if known: summary.append(f"I see {len(faces)} people including {', '.join(known)}") else: summary.append(f"I see {len(faces)} people I don't recognize") else: summary.append("I don't see anyone") # Scene objects = obs.get('objects', {}) complexity = objects.get('scene_complexity', 'unknown') summary.append(f"The scene is {complexity}") # Colors color = obs.get('dominant_color', 'unknown') summary.append(f"with {color} tones") return ". ".join(summary) + "." def release(self): """Release camera""" self.camera.release() if __name__ == '__main__': print("Testing Eden's vision system...\n") vision = EdenVision() print("\n1. Capturing and analyzing frame...") obs = vision.perceive(include_text=True) print("\n2. What Eden sees:") print(json.dumps(obs, indent=2, default=str)) print("\n3. Visual summary:") print(vision.get_visual_summary()) print("\n4. Saving test image...") frame = vision.capture_frame() if frame is not None: cv2.imwrite('/tmp/eden_vision_test.jpg', frame) print(" Saved to /tmp/eden_vision_test.jpg") vision.release() print("\n✅ Vision test complete!")