Color images are captured by pheromone traps installed at multiple locations in a field with vegetable crops (tomato) at Volos, Greece by a commercial provider of pheromone-based pest control solutions. The pheromone attracted the pest of interest into the trap where they became stuck to the adhesive surface. A digital camera (Svpro 13MP, sensor: Sony 1/3” IMX213) was used to capture the trapped insects. The resolution of images taken was 3840x2160. The working distance was set to 20 cm based on the camera specifications, in order to obtain a large enough field of view to capture the whole adhesive board trap. Using that equipment, 225 images were captured in indoor and outdoor conditions and chosen to represent the insect variability. For image annotation, Roboflow was used. The annotation process aims to label the location and class of the insect pests in the image. Roboflow was chosen because it can generate various annotations formats for different object detection models. The whole process was carried out by an experienced entomologist. Two insects were labelled, whiteflies, and black aphids. In total 5904 insect instances were labelled, out of which 2431 are whiteflies and 3473 are black aphids. Note that there are 26.24 annotations per image at average across the two classes. For validating the Yellow Glue Paper Traps Dataset, we split the whole images into training and validation subsets. In total, the dataset was split into 180 images for training and 45 for validation with 80-20% proportion. The split was chosen so as to keep a similar ratio and the classes which would ensure same distribution to training and validation subsets. The insect count in each subset can be seen in the following table: Dataset splits insect counts Dataset split Images Insect Instances Whiteflies Black Aphids Training 180 4773 1926 2847 Validation 45 1131 505 626 Total 225 5904 2431 3473