Dataset and Benchmarking of Real-Time Embedded Object Detection for RoboCup SSL

When producing a model to object detection in specific context, the first obstacle is have dataset labeling desired classes. In RoboCup, some leagues already more than one train and evaluate model. However, Small Size League (SSL), there not such available yet. This paper presents an open-source be used as benchmark for real-time SSL. work also presented pipeline train, deploy, Convolutional Neural Networks (CNNs) models low-power embedded system. proposed with state-of-art optimized models. this dataset, MobileNet SSD v1 achieves $$44.88\%$$ AP ( $$68.81\%$$ $$_{50}$$ ) at 94 Frames Per Second (FPS), while running on SSL robot.