Identifying and quantifying weeds is a crucial aspect of agriculture for efficiently controlling them. Weeds compete with the crop for nutrients, minerals, physical space, sunlight, and water, causing problems in crops ranging from low production to economic losses and environmental deterioration of the land. Weed quantification is generally a manual process requiring significant time and precision. Convolutional Neural Networks (CNN) are very common in weed quantification. Thus, the purpose of this research is the adaptation of the ResNeXt50 CNN architecture for semantic segmentation tasks, focused on the automatic quantification of weeds (Broadleaf dock, Dandelion, Kikuyo grass, and other unidentified classes) in potato fields using RGB images acquired by the DJI Mavic 2 Pro drone. The analytical model was trained following the Knowledge Discovery in Databases (KDD) methodology using Python and the TensorFlow-Keras frameworks. The results indicate that the modified ResNeXt50 model presented a mean IoU of 0.7350, a performance comparable to the values reported by other authors considering fewer weed classes. The Student´s t-test and Pearson correlation coefficient were applied to contrast the weed coverage from the model predictions and the ground truth, indicating no statistically significant differences between both measurements in most weed classes.

Today, cars are an indispensable element in the society, as well as the vehicle diagnosis of minor and serious mechanical failures. This process is carried out through two methods: (i) manually, inspecting possible common causes; and (ii) automatically, using a failure identification scanner. In both cases the assistance of a car expert is required. However, how could a common user briefly diagnose vehicle failures? The objective of this project has been to build an expert system module for vehicular diagnosis to help the common user, identifying automotive failures and the severity of the vehicle damage. It also helps to prevent major damages and possible accidents, as well as to achieve a technical and effective communication when the situation is being explained to the mechanical assistance which can be even by telephone. The module design was composed by four phases: (i) do background research about failure diagnosis, (ii) production rules; (iii) inference engine; and (iv) user interface. The results show that the expert system module is 71,43% effective, so that it helps the common user to identify electronic engine failures without the assistance of a professional in the area.