 The paper proposes a novel IoT fog computing equipped robotic system for the accurate identification of weeds and soybeans in a hazy environment. The system uses a two-dimensional convolutional neural network, 2D, CNN, based deep learning approach to classify the images into four categories, soil, soybean, grass, and weeds. The proposed system is connected to the internet via an IoT enabled server, which enables real-time data processing and classification. Additionally, the system is equipped with edge-based fog computing, which enhances its reliability and performance. The system was tested using a dataset of 150 by 150 pixels with three channels. The results showed that the proposed system achieved a high accuracy of 97%, demonstrating its effectiveness in identifying weeds and soybeans in a hazy environment. This article was authored by Kansal Isha, Kuler Vikas, Varma Jodi, and others.