This study compares the performance of three YOLO-based object detection models—YOLOv3, YOLOv5, and YOLOv8—for vehicle detection and classification at an urban intersection in Montería, Colombia. Recordings from five consecutive days, spanning three time slots, were used, totaling approximately 135,000 frames with variability in lighting and weather conditions. Frames were preprocessed by maintaining the aspect ratio and were normalized according to each model. The evaluation employed models pre-trained on COCO, without fine-tuning, enabling an objective assessment of their generalization capacity. Precision, recall, F1-score, and mAP@0.5 were computed globally and by vehicle class. YOLOv5 achieved the best balance between precision and recall (F1-score = 0.78) and the highest mAP (0.63), while YOLOv3 showed lower recall and mAP, and YOLOv8 performed competitively but slightly below YOLOv5. Cars and motorcycles were the most robust classes, whereas bicycles and trucks showed greater detection challenges. Visual evaluation confirmed stable performance on cloudy days and in light rain, with reduced accuracy under sunny conditions with high contrast. These findings highlight the potential of modern YOLO architectures for intelligent urban traffic monitoring and management. The generated dataset constitutes a replicable resource for future mobility research in similar contexts.

Near-infrared (NIR) imaging plays a crucial role in precision agriculture; however, the high cost of multispectral sensors limits its widespread adoption. In this study, we generate synthetic NIR images (2592 × 1944 pixels) of pineapple crops from standard RGB drone imagery using the Pix2PixHD framework. The model was trained for 580 epochs, saving the first model after epoch 1 and then every 10 epochs thereafter. While models trained beyond epoch 460 achieved marginally higher metrics, they introduced visible artifacts. Model 410 was identified as the most effective, offering consistent quantitative performance while producing artifact-free results. Evaluation of Model 410 across 229 test images showed a mean SSIM of 0.6873, PSNR of 29.92, RMSE of 8.146, and PCC of 0.6565, indicating moderate to high structural similarity and reliable spectral accuracy of the synthetic NIR data. The proposed approach demonstrates that reliable NIR information can be obtained without expensive multispectral equipment, reducing costs and enhancing accessibility for farmers. By enabling advanced tasks such as vegetation segmentation and crop health monitoring, this work highlights the potential of deep learning–based image translation to support sustainable and data-driven agricultural practices. Future directions include extending the method to other crops, environmental conditions and real-time drone monitoring.