<p>Recent advances have integrated computer vision with supervised deep learning to optimize manufacturing processes. Common applications include in situ process control, anomaly detection, and quality control. Electrospinning produces micro- and nano-scale fibers from polymer substrates in a high-voltage electric field. Its versatile configuration enables the rapid fabrication of diverse structures using various materials for a broad range of applications. Regardless of the final morphology of electrospun fibers, the formation of a charged polymer jet under the interaction between electrostatic forces and surface tension is initiated by deforming the droplet into a conical shape known as the Taylor cone. Undesired phenomena in the Taylor cone, such as multi-jet formation and jet blockage, largely compromise the precision of the fabrication process. Therefore, the prerequisite for achieving precise process control in electrospinning is to stabilize the Taylor cone. However, there is limited research on real-time image analysis of the Taylor cone. To address this challenge, a convolutional long short-term memory deep learning model has been developed. This model is trained to classify different types of Taylor cones (single jet, no jet, jet Buildup, and jet branching) from high-speed camera images. The optimized model achieved a high prediction accuracy of 99% on the test set. Several live Taylor cone attenuation experiments were conducted to assess the model's ability to detect transitions in real time. The live predictions were then compared with the final fiber properties, such as tensile strength, fiber density, and fiber diameter uniformity, with relevant statistical analyses performed. The model's robustness was validated through these experimental tests involving continuous monitoring for anomalies. The developed temporal anomaly detection system was also used to characterize fiber diameter variability, inter-fiber distance, and mechanical tensile properties as a function of the anomaly percentage detected during a certain period of operation. This study sets the foundation for applying deep learning-based anomaly detection in precision electrospinning, paving the way for accurate fiber morphology prediction and automated process optimization driven by artificial intelligence.</p>