The omnipresence of Wi-Fi-enabled Internet of Things (IoT) networks has enveloped our surroundings, facilitating remote monitoring through controllers equipped with integrated Wi-Fi capabilities that leverage sensor data analysis. However, deploying sensors and actuators within these networks often entails substantial installation costs and spatial requirements. One particularly promising application of such networks is the utilization of Wi-Fi Channel State Information (CSI) to discern and categorize human activities, offering manifold benefits across diverse domains. In the realm of healthcare, this technology proves invaluable in identifying instances of falls, anomalous movements, or other aberrant activities, promptly notifying caregivers of potential medical crises. Similarly, surveilling individuals traversing restricted zones in the security sphere becomes more sophisticated and efficient. Nonetheless, analyzing Wi-Fi CSI data presents an intricate challenge, characterized by inherent noise and complex patterns that necessitate advanced Machine Learning (ML) or Deep Learning (DL) techniques for accurate interpretation. This research endeavors to address this challenge by focusing on the development and application of deep learning models tailored to a dataset generously provided by Silicon Labs Inc., with the ultimate aim of effectively classifying a spectrum of human activities-ranging from sitting, standing, falling, walking to running-based on CSI data collected via the ESP-32 board.