A Solution to Product detection in Densely Packed Scenes

TL;DR

This paper presents a modified Cascade R-CNN approach with a novel random crop strategy and hyper-parameter optimization to improve product detection in densely packed scenes, achieving a 58.7 mAP on SKU-110k.

Contribution

It introduces a new random crop method and detailed analysis of detector stages to enhance detection performance in densely packed scenes.

Findings

Achieved 58.7 mAP on SKU-110k dataset.

Identified bottlenecks in detector stages affecting performance.

Optimized hyper-parameters and sampling strategies for better results.

Abstract

This work is a solution to densely packed scenes dataset SKU-110k. Our work is modified from Cascade R-CNN. To solve the problem, we proposed a random crop strategy to ensure both the sampling rate and input scale is relatively sufficient as a contrast to the regular random crop. And we adopted some of trick and optimized the hyper-parameters. To grasp the essential feature of the densely packed scenes, we analysis the stages of a detector and investigate the bottleneck which limits the performance. As a result, our method obtains 58.7 mAP on test set of SKU-110k.