Dilated-Scale-Aware Attention ConvNet For Multi-Class Object Counting

TL;DR

This paper introduces a multi-class object counting network that uses point-level annotations and a multi-mask structure to improve accuracy across multiple object categories, achieving state-of-the-art results.

Contribution

It presents a novel multi-class counting network with a multi-mask structure that effectively suppresses feature interference among categories, using only point annotations.

Findings

Achieves state-of-the-art performance on benchmark datasets.

Effectively handles multiple object categories in counting tasks.

Uses point-level annotations instead of bounding boxes.

Abstract

Object counting aims to estimate the number of objects in images. The leading counting approaches focus on the single category counting task and achieve impressive performance. Note that there are multiple categories of objects in real scenes. Multi-class object counting expands the scope of application of object counting task. The multi-target detection task can achieve multi-class object counting in some scenarios. However, it requires the dataset annotated with bounding boxes. Compared with the point annotations in mainstream object counting issues, the coordinate box-level annotations are more difficult to obtain. In this paper, we propose a simple yet efficient counting network based on point-level annotations. Specifically, we first change the traditional output channel from one to the number of categories to achieve multiclass counting. Since all categories of objects use the same feature extractor in our proposed framework, their features will interfere mutually in the shared feature space. We further design a multi-mask structure to suppress harmful interaction among objects. Extensive experiments on the challenging benchmarks illustrate that the proposed method achieves state-of-the-art counting performance.