Weed infestations significantly reduce rice productivity and grain quality, necessitating early and efficient weed detection methods for effective smart farming management. Weeds are major problem in rice farming, competing with rice for essential resources such as water, light, nutrients, and space, potentially resulting in yield losses to a maximum extent if not adequately managed. This paper reviews recent advances and challenges in automatic weed detection using remote sensing technologies and artificial intelligence (AI), with a focus on site-specific weed management. The paper discusses the imaging sensors commonly used in weed detection, including RGB, multispectral, and hyperspectral sensors, highlighting their potential applications, advantages, and drawbacks. Various types of image features-spectral, color model, shape, and texture are explored for their efficacy in distinguishing weeds from crops. The comparison of different remote sensing platforms, including ground-based, UAVs, and satellites is evaluated, emphasizing UAVs’ cost-effectiveness and flexibility. Machine learning and deep learning techniques employed to differentiate weed species from crops are reviewed. Results indicated that the AI approaches have effectively generated highly accurate maps for identifying weeds and the selective application of herbicide. Although remote sensing and AI have been explored for weed detection in other crops, their application in paddy fields remains in the early stages. This approach has significant potential for improving weed management economically and could be expanded in smart farming for paddy fields.

Table 1: Summary of the key differences between RGB, multispectral, and hyperspectral sensors

Reference: Islam, M.M., Khairunniza-Bejo, S., Hamdani, M.S.A. et al. Automatic weed detection in paddy field: a review. J Plant Dis Prot 133, 19 (2026).

Link:  https://doi.org/

Date of Input: 31/03/2026 | Updated: 31/03/2026 | ainzubaidah