Global Wheat Head Detection (GWHD) Dataset: A Large and Diverse Dataset of High-Resolution RGB-Labelled Images to Develop and Benchmark Wheat Head Detection Methods

Abstract

The detection of wheat heads in plant images is an important task for estimating pertinent wheat traits including head population density and head characteristics such as health, size, maturity stage, and the presence of awns. Several studies have developed methods for wheat head detection from high-resolution RGB imagery based on machine learning algorithms. However, these methods have generally been calibrated and validated on limited datasets. High variability in observational conditions, genotypic differences, development stages, and head orientation makes wheat head detection a challenge for computer vision. Further, possible blurring due to motion or wind and overlap between heads for dense populations make this task even more complex. Through a joint international collaborative effort, we have built a large, diverse, and well-labelled dataset of wheat images, called the Global Wheat Head Detection (GWHD) dataset. It contains 4700 high-resolution RGB images and 190000 labelled wheat heads collected from several countries around the world at different growth stages with a wide range of genotypes. Guidelines for image acquisition, associating minimum metadata to respect FAIR principles, and consistent head labelling methods are proposed when developing new head detection datasets. The GWHD dataset is publicly available at http://www.global-wheat.com/and aimed at developing and benchmarking methods for wheat head detection.

Figure 1

Overview of the harmonization process conducted. Images were first rescaled using bilinear interpolation up- or downsampling techniques. Then, the rescaled images were split into 1024 × 1024 squared patches.

Figure 2

Examples of wheat heads difficult to label. These examples are zoomed-in views from images contained in the dataset, with different zoom factors. It includes overlapping heads (a–c), heads at emergence (d), heads that are partly cut at the border of the image (e), and images with a low illumination (f). Note that image (d) was removed from the dataset because of the ambiguity of heads at emergence. Wheat heads in the image (e) were not labelled because less than 30% of their basal part is visible, as defined in Section 2.4.

Figure 3

Distribution of the number of bounding boxes per image (a) and bounding boxes size^∗ (b) in the GWHD dataset. ^∗The bounding box size is defined as the square root of the bounding box area in pixel.

Figure 4

Example of images from different acquisition sites after cropping and rescaling.

Figure 5

A selection of bounding boxes for each sub-dataset. The same size of pixels is used across all the bounding boxes displayed.

Figure 6

Comparison of GWHD dataset with other object detection datasets. Both axes are in log-scale.