Complete Python 3.12 migration and full ML environment setup

- Migrate from Python 3.13 to 3.12.5 using pyenv
- Install complete ML dependencies for Apple Silicon:
  * TensorFlow 2.16.2 with tensorflow-macos and tensorflow-metal
  * PyTorch 2.7.1 with Apple Silicon optimization
  * OpenCV 4.11.0 for computer vision
  * Ultralytics 8.3.159 for YOLO models
  * Librosa 0.11.0 for audio processing
  * scikit-learn 1.7.0 for machine learning
- Add ffmpeg-python for media processing
- Update config.yaml for local macOS development
- Create required directories (models, data, logs)
- Verify GPU support: TensorFlow Metal enabled
- Confirm both minimal and full versions working

Status:
✅ Python 3.12.5 (full TensorFlow compatibility)
✅ Apple Silicon GPU acceleration
✅ Complete ML pipeline ready
✅ All dependencies resolved

🤖 Generated with [Claude Code](https://claude.ai/code)

Co-Authored-By: Claude <noreply@anthropic.com>

Author: atsuki-ichikawa

.python-version

@@ -0,0 +1 @@
+3.12.5

config.yaml

@@ -1,18 +1,19 @@
 # Bird Monitoring Pipeline Configuration
+# macOS Local Development Settings
 
 # Environment settings
-environment: production
-project_root: /app
+environment: development
+project_root: .
 
 # Directories
-models_dir: /app/models
-data_dir: /app/data
-logs_dir: /app/logs
+models_dir: models
+data_dir: data
+logs_dir: logs
 
 # Model paths (update these when models are available)
-birdnet_model_path: /app/models/birdnet/model.tflite
-yolo_model_path: /app/models/yolo/yolov10n.pt
-transfg_model_path: /app/models/transfg/model.pth
+birdnet_model_path: null
+yolo_model_path: null
+transfg_model_path: null
 
 # Audio processing settings
 audio_segment_length: 3.0
@@ -37,15 +38,15 @@ video_only_base_confidence: 0.4
 audio_video_base_confidence: 0.7
 
 # Web application
-web_host: 0.0.0.0
+web_host: localhost
 web_port: 8000
-web_reload: false
+web_reload: true
 
 # Logging
 log_level: INFO
 log_format: "%(asctime)s - %(name)s - %(levelname)s - %(message)s"
 
 # Performance
 max_workers: 4
-batch_size: 32
-gpu_enabled: false
+batch_size: 16
+gpu_enabled: true