The competition of nutrients caused by weeds in the agricultural production has caused significant losses in the development of agricultural crops at the same time as there is a growing environmental concern regarding the use of agrochemicals in the field.

As the agro-industrial sector employs agricultural systems based on minimum planting with increasing dependence on herbicides, it is essential the creation of innovative proposals for weed management strategies based on precision spray technologies. The objective of this work is to evaluate the capacity of weed identification by an active optical sensor in relation to target size and surface type.

The active sensor WeedIT (Rometron B.V., Steenderen - Netherlands) was used attached to an agricultural vehicle. The sensors were spaced 1.0 m from the stand and operated at a height of 1.10 m from the ground. A sensor can control five solenoids each with spacing of 0.20 m. It was used a hydraulic pump with flow rate of 18,0 L min^-1 (Pressure: 5 bar - 72psi), and speed sensor that aided in the correct operation of the equipment and also in the acquisition of experimental data.

Five different sizes of circular targets (8.81 cm², 3.77 cm², 2.38 cm², 1.11 cm², 0.49 cm²) were determined on eight lanes of 6.0 m with targets spaced at 40.0 cm, totalising 24 targets of each size. The evaluations were performed with four preset levels pre-defined by the manufacturer at an average speed of 10.0 km h^-1 (2.78 m s^-1). The experiment was conducted on three different application surfaces: asphalt, exposed soil and soil covered by straw. Sensor identification was obtained by means of digital images from a digital camera installed at the rear of the agricultural vehicle. From the edition of the images were determined the percentages of targets identified according to the size of the target and type of surface.

It was observed that with the change from preset number 4 (less sensitive) to 1 (more sensitive) it is possible to verify an increasing rate of target identification on all evaluated surfaces, but this increasing rate was more abrupt on the asphalt surface.

The manufacturer indicates the possibility of identifying targets with up to 1.5 cm² and it was observed that this occurs if the most sensitive preset is used, even though there were no total identification of the targets with more than 1.5 cm². Another point observed was the interference of vegetal residues in both the exposed soil and straw, being this effect potentiated in more sensitive presets.