CNN 모델의 효율적 배포를 위한 양자화 학습 전략 비교 및 TFLite 변환 영향 분석 Comparison of Quantization-Aware Training Strategies and Analysis of TFLite Conversion Effects for Efficient Deployment of CNN Models