The growing need for precision agriculture contributes to integrating advanced technology to enhance agricultural yield and sustainability. This research presents an advanced intelligent robotic crop replanting and recovery system, incorporating Dense Convolutional Networks (DenseNet) and Edge Computing. The system is designed to function autonomously in agricultural settings, offering the ability to analyze and make decisions in real-time. The DenseNet architecture, known for effectively transmitting features and mitigating the vanishing-gradient issue, is used to analyze complex multispectral images obtained by the robotic system. Using this deep learning (DL) technique allows for the precise identification of failing crops, enabling the detection of regions that need replanting or recovery with high accuracy. By integrating Edge Computing, data processing moves close to the source, resulting in a substantial reduction in latency and enabling quick action to be taken in the field. The proposed approach utilizes the advantages of DenseNet and Edge Computing to improve the efficiency of crop management operations, reduce resource waste, and increase crop output. This method is advantageous in agricultural operations of significant size, where quick and accurate interventions are crucial for preserving crop health and yield. The findings indicate that the system can completely transform traditional agricultural methods, providing a scalable solution for the sustainable and effective management of crops in the era of intelligent farming. © 2025 Elsevier B.V., All rights reserved.