Wheat Spike Detection and Counting in the Field Based on SpikeRetinaNet

Abstract

The number of wheat spikes per unit area is one of the most important agronomic traits associated with wheat yield. However, quick and accurate detection for the counting of wheat spikes faces persistent challenges due to the complexity of wheat field conditions. This work has trained a RetinaNet (SpikeRetinaNet) based on several optimizations to detect and count wheat spikes efficiently. This RetinaNet consists of several improvements. First, a weighted bidirectional feature pyramid network (BiFPN) was introduced into the feature pyramid network (FPN) of RetinaNet, which could fuse multiscale features to recognize wheat spikes in different varieties and complicated environments. Then, to detect objects more efficiently, focal loss and attention modules were added. Finally, soft non-maximum suppression (Soft-NMS) was used to solve the occlusion problem. Based on these improvements, the new network detector was created and tested on the Global Wheat Head Detection (GWHD) dataset supplemented with wheat-wheatgrass spike detection (WSD) images. The WSD images were supplemented with new varieties of wheat, which makes the mixed dataset richer in species. The method of this study achieved 0.9262 for mAP50, which improved by 5.59, 49.06, 2.79, 1.35, and 7.26% compared to the state-of-the-art RetinaNet, single-shot multiBox detector (SSD), You Only Look Once version3 (Yolov3), You Only Look Once version4 (Yolov4), and faster region-based convolutional neural network (Faster-RCNN), respectively. In addition, the counting accuracy reached 0.9288, which was improved from other methods as well. Our implementation code and partial validation data are available at https://github.com/wujians122/The-Wheat-Spikes-Detecting-and-Counting.

Improved Detection and Analysis of Wheat Spikes U sing Multi-Stage Convolutional Neural Network

“…We applied the Cascade R-CNN architecture [10] for the detection by fine-tuning the hyper parameters and customizing the model to suit the need of our spike detection task. In doing so, we observed an improvement over existing state-of-the-art methods developed by Hasan et al [3] and Wen et al [7]; a 2.9% and and 11.2% increase in Mean Average Precision (mAP), 8% and 13% increase in Recall, 9% and 15% increase in Accuracy metric as well as 5% and 9% increase in Average F1 score for detection count. Moreover, the experimental results demonstrate its significance in various analysis tasks such as yield estimation, plant growth measure, genotype traits, etc.…”

“…In our study, we intended to develop a detection method that can classify wheat spikes with better accuracy. We have compared our results with two existing approaches developed by hasan et al [3] wen et al [7] and claim to be more efficient, feasible, and robust in the field of spike detection in a complex environment.…”

“…Many deep neural networks, Fast R-CNN, Faster R-CNNN, RetinaNet, YOLOv5 (You Only Look Once), etc. are developed for wheat spike detection and counting [3,7,8]. These models are trained with a large spike dataset so that later it would be able to detect spikes from images.…”

“…Wen et al [7] developed a technique based on Spik-eRetinaNet in order to detect and count small dense objects in complex scenes. SpikeRetinaNet is an opti-mized version of the RetinaNet model and is consisting of three crucial steps: the usage of BiFPN for more effective integration of multiscale features, DSA block for improved network refinement, and the application of Soft-NMS to solve the occlusion problem.…”

Detection method of wheat spike improved YOLOv5s based on the attention mechanism

“…Wheat is an important food crop in our country. In 2021, the planting area of wheat will be 22.911 million hectares, and the output will be 134 million tons in our country; China is the largest wheat producer in the world ( Sreenivasulu and Schnurbusch, 2012 ; Ge et al, 2018 ; Chen et al, 2021 ; Wen et al, 2022 ). However, the current COVID-19 epidemic is raging, the domestic and foreign environments are complex and changeable, abnormal weather and natural disasters are frequent, and food security is facing severe challenges ( Laborde et al, 2020 ; FAO, 2021 ; Ministry of Emergency Management of the People’s Republic of China, 2022 ).…”