IP102: A Large-Scale Benchmark Dataset for Insect Pest Recognition

Xiaoping Wu¹, Chi Zhan¹, Yu-Kun Lai², Ming-Ming Cheng¹, Jufeng Yang¹∗

¹Nankai University     ²Cardiff University

Abstract

Insect pests are one of the main factors affecting agricultural product yield. Accurate recognition of insect pests facilitates timely preventive measures to avoid economic losses. However, the existing datasets for the visual classification task mainly focus on common objects, e.g., flowers and dogs. This limits the application of powerful deep learning technology on specific domains like the agricultural field. In this paper, we collect a large-scale dataset named IP102 for insect pest recognition. Specifically, it contains more than 75,000 images belonging to 102 categories, which exhibit a natural long-tailed distribution. In addition, we annotate about 19,000 images with bounding boxes for object detection. The IP102 has a hierarchical taxonomy and the insect pests which mainly affect one specific agricultural product are grouped into the same upper-level category. Furthermore, we perform several baseline experiments on the IP102 dataset, including handcrafted and deep feature based classification methods. Experimental results show that this dataset has the challenges of inter- and intra-class variance and data imbalance. We believe our IP102 will facilitate future research on practical insect pest control, fine-grained visual classification, and imbalanced learning fields. We make the dataset and pre-trained models publicly available at https://github.com/xpwu95/IP102.

Paper

IP102: A Large-Scale Benchmark Dataset for Insect Pest Recognition. Xiaoping Wu, Chi Zhan, Yu-Kun Lai, Ming-Ming Cheng, and Jufeng Yang. CVPR, 2019, Oral. [pdf] [bib] [poster] [video] [Project Page] [Dataset]

Highlights

• The largest public dataset for insect pest recognition. This dataset contains 102 insect pests, including 75,222 images with category labels and 18,976 images with bounding boxes.
• Extensive experiments on the proposed dataset.

Motivation

• Insect pest is one of the main factors affecting agricultural product yield. Accurate recognition of insect pests facilitates timely preventive measures to avoid economic losses.
• Existing small-scale insect pest datasets cannot well satisfy the requirement of deep technology.

Statistics of the proposed IP102

Figure 1: Statistics of the proposed IP102 dataset. (a) Hierarchical taxonomy system. (b) Statistical information.

Challenges of the proposed IP102

Figure 2: Challenges of the proposed IP102 dataset. (a) Imbalanced distribution. (b) Intra- & inter-class variance.

Benchmark Experiments

Classification performance of handcrafted and deep features

Classification performance with different hierarchical labels

Detection performance

More examples