With a rapidly growing global population, meeting increasing agricultural demands has become imperative. Smart farming, powered by machine learning, has the potential to address this issue but faces hurdles in managing big data with high velocity. In this study, we implement a comprehensive big data pipeline for cotton bloom detection that utilizes Azure cloud computing resources and employs YOLOv5 for real-time and batch processing. The model achieves a high mean Average Precision (mAP) score of 0.96 for cotton bloom classification using 2021 data. We also explore the use of Principal Component Analysis (PCA) as a data compression method to optimize pipeline execution time and storage space. Rigorously tested for scalability on distinct 2022 data, our pipeline incorporates downsampling with masking as an effective pre-processing step to reduce computational overhead while preserving accuracy. This research underscores the potential of cloud computing in driving efficient big data processing in precision agriculture, enabling accurate crop yield prediction through advanced plant phenotyping techniques.