Search papers, labs, and topics across Lattice.
This study investigates the effectiveness of synthetic and derived images in enhancing a YOLOv8n model for detecting waste disposal on campus, utilizing a dataset of 148 real photographs. Through twelve joint-training configurations and multiple seed evaluations, the research found that while certain augmentation strategies like isolated-object images improved performance, none surpassed the baseline accuracy achieved with real images alone. The findings indicate that despite the potential of synthetic data, the small size of the test set limits the ability to draw definitive conclusions about class-specific performance improvements.
Despite the promise of synthetic data, the YOLOv8n model's performance remained unchanged, revealing that real images are still king in waste detection tasks.
Incorrect disposal can contaminate campus recycling streams, and a bin-mounted camera could provide feedback as an item is discarded. We evaluated whether synthetic and derived images improve a YOLOv8n detector for this view. The real dataset contained 148 campus photographs: 86 for training, 31 for validation, and 31 for testing. Twelve joint-training configurations varied the amount and source of added images. We repeated seven principal settings with four matched seeds and computed bootstrap percentile intervals over those seeds. The real-only model reached a mean mAP@0.5 of 0.691 [0.665, 0.722]. Background replacement reduced the mean to 0.560 [0.499, 0.619], isolated-object images gave 0.680 [0.644, 0.724], and the full augmentation pool gave 0.487 [0.438, 0.537]. We also tested hand-and-forearm composites because every real photo showed a held object. Two cutouts in the initial composite set came from test photographs, so we discarded that experiment, rebuilt the set with training-split cutouts, and reran all four seeds. The corrected paired difference was +0.034 [-0.063, 0.199], which does not support a reliable hand-composite effect. Single-seed transfer experiments produced source-dependent rankings between joint mixing and sequential pretraining. None of the evaluated configurations exceeded the real-only baseline. The reported intervals quantify seed variation; the 31-photo test set remains too small for strong class-specific conclusions.
