Insect pests significantly affect tomato crops, causing yield loss and quality degradation. Thrips, whiteflies, and Tuta absoluta are key pests managed using sticky traps, yet traditional counting methods are laborious and error-prone. Integrated Pest Management (IPM) aims to enhance pest control and reduce pesticide use. Advances in artificial intelligence (AI) enable automatic pest detection and counting. This study presents a real-time system using computer vision and deep learning, specifically the YOLOv8 model and Slicing Aided Hyper Inference (SAHI) method. The system was tested on three datasets for thrips, whiteflies, and Tuta absoluta. Transfer learning and data augmentation were employed to improve performance with limited data. Experiments showed high accuracy, with YOLOv8 achieving an mAP@0.50 of 50% for thrips, 60% for whiteflies, and 90% for Tuta absoluta, and mAP@0.50–0.95 of 20%, 40%, and 80%, respectively. The model proved robust across various conditions and pest densities, supporting timely pest control interventions. The system enhances pest monitoring accuracy and efficiency, reduces manual counting, and supports sustainable agriculture by minimizing pesticide use. It is scalable for different agricultural settings and pest types, contributing to better crop protection and sustainable practices.