-This dataset provides high-resolution RGB images of wheat–weed interactions acquired under controlled mesocosm conditions, ensuring high reproducibility and limited environmental variability. -The data cover key wheat development stages from BBCH 12 (2-leaf stage) to BBCH 67 (anthesis), based on 11 acquisition dates, enabling the analysis of temporal dynamics in crop–weed competition. -The inclusion of multiple experimental treatments (wheat monoculture, wheat with one weed species, and wheat with two weed species combining contrasting competitiveness) enables the study of interspecific interactions and competition effects. -Four weed species with contrasting functional traits, densities, and competitive abilities are included, providing a relevant gradient for ecological and agronomic analyses. -The high spatial resolution (0.3 mm·px⁻¹) and top-down imaging configuration facilitate the development and evaluation of computer vision and machine learning approaches for plant detection, segmentation, and classification. -This dataset is relevant for research in agroecology, precision agriculture, and plant phenotyping, and can serve as a reference dataset for developing and validating image-based methods under controlled conditions.

RGB images were acquired using a Canon EOS 760D camera positioned 2 m above the mesocosms (nadir view) at 11 dates during the wheat growth cycle betwwen BBCH12 (begining tillering) to BBCH67 (anthesis)