Abstract

In hazy environments, the computer vision system may require to perform object detection. The performance of the object detection methods degrades in a hazy environment. To overcome this issue, we propose a Bi-stream feature fusion (BFF) network for object detection in a hazy environment. The BFF network consists of three modules: hybrid input, Bi-stream feature extractor (BFE), and multi-level feature fusion. We present the notion of hybrid input to extract features from the hazy images in an effective manner. This paper leverages the hybrid input for feature extraction from the hazy images to avoid the requirement of enhancement in hazy object detection. The proposed BFE network extracts multi-level features from the hazy image and hybrid input. The multi-level feature fusion (MFF) network performs the convolution-based adaptive feature fusion and processes the extracted features. The proposed BFF model outperforms other state-of-the-art methods in hazy environments while achieving competitive performance in normal conditions. Another challenge in hazy object detection is the unavailability of a dataset with sufficient samples and classes. In this work, we developed a synthetic object detection dataset for a hazy environment (DHOD). The DHOD dataset contains twenty object classes with more than twenty thousand samples.

中文翻译：

---

Bff：雾霾环境下目标检测的双流特征融合

摘要

在雾霾环境中，计算机视觉系统可能需要执行物体检测。物体检测方法的性能在雾霾环境中会下降。为了克服这个问题，我们提出了一种双流特征融合（BFF）网络，用于雾霾环境中的目标检测。 BFF网络由三个模块组成：混合输入、双流特征提取器（BFE）和多级特征融合。我们提出了混合输入的概念，以有效的方式从模糊图像中提取特征。本文利用混合输入从模糊图像中提取特征，以避免模糊目标检测中增强的要求。所提出的 BFE 网络从模糊图像和混合输入中提取多级特征。多级特征融合（MFF）网络执行基于卷积的自适应特征融合并处理提取的特征。所提出的 BFF 模型在雾霾环境中优于其他最先进的方法，同时在正常条件下实现有竞争力的性能。模糊对象检测的另一个挑战是缺乏具有足够样本和类别的数据集。在这项工作中，我们开发了一个用于雾霾环境（DHOD）的合成对象检测数据集。 DHOD 数据集包含 20 个对象类别，超过两万个样本。