Ensemble-based Adaptive Single-shot Multi-box Detector

TL;DR

This paper enhances the SSD object detection method by introducing adaptive default box selection based on data distribution and an ensemble approach, significantly improving accuracy especially on small datasets with complex objects.

Contribution

It presents a novel adaptive box selection method and an ensemble strategy for SSD, leading to notable performance gains over the standard SSD.

Findings

Adaptive box selection improves mAP by 3%.

Ensemble SSD improves mAP by 8%.

Performance gains are significant for small and complex object datasets.

Abstract

We propose two improvements to the SSD---single shot multibox detector. First, we propose an adaptive approach for default box selection in SSD. This uses data to reduce the uncertainty in the selection of best aspect ratios for the default boxes and improves performance of SSD for datasets containing small and complex objects (e.g., equipments at construction sites). We do so by finding the distribution of aspect ratios of the given training dataset, and then choosing representative values. Secondly, we propose an ensemble algorithm, using SSD as components, which improves the performance of SSD, especially for small amount of training datasets. Compared to the conventional SSD algorithm, adaptive box selection improves mean average precision by 3%, while ensemble-based SSD improves it by 8%.