Abstract:High-throughput phenotyping (HTP) of seeds, also known as seed phenotyping, is the comprehensive assessment of complex seed traits such as growth, development, tolerance, resistance, ecology, yield, and the measurement of parameters that form more complex traits. One of the key aspects of seed phenotyping is cereal yield estimation that the seed production industry relies upon to conduct their business. While mechanized seed kernel counters are available in the market currently, they are often priced high and sometimes outside the range of small scale seed production firms' affordability. The development of object tracking neural network models such as You Only Look Once (YOLO) enables computer scientists to design algorithms that can estimate cereal yield inexpensively. The key bottleneck with neural network models is that they require a plethora of labelled training data before they can be put to task. We demonstrate that the use of synthetic imagery serves as a feasible substitute to train neural networks for object tracking that includes the tasks of object classification and detection. Furthermore, we propose a seed kernel counter that uses a low-cost mechanical hopper, trained YOLOv8 neural network model, and object tracking algorithms on StrongSORT and ByteTrack to estimate cereal yield from videos. The experiment yields a seed kernel count with an accuracy of 95.2\% and 93.2\% for Soy and Wheat respectively using the StrongSORT algorithm, and an accuray of 96.8\% and 92.4\% for Soy and Wheat respectively using the ByteTrack algorithm.

What problem does this paper attempt to address?

### Problems the Paper Aims to Solve This paper aims to address a critical issue in the seed packaging industry related to High-Throughput Phenotyping (HTP): grain yield estimation. Specifically, the paper attempts to develop a low-cost, high-precision seed counting method to replace the expensive mechanized seed counting equipment available in the market. ### Background and Motivation 1. **Industry Demand**: Seed packaging companies typically rely on accurate seed quantity estimates to conduct their business. However, existing mechanized seed counting equipment is expensive and unaffordable for small seed packaging enterprises. 2. **Existing Problems**: Currently, many small seed packaging companies can only package seeds by weight, but the weight of seeds is greatly affected by environmental factors, leading to discrepancies between the actual number of seeds and the expected number. 3. **Technical Challenges**: Traditional image processing algorithms have limitations in terms of video frame rate sensitivity, and existing commercial seed counting equipment is costly, limiting its application in small enterprises. ### Solution 1. **Hardware Design**: The paper proposes a low-cost hardware device, including a mechanical funnel and a mobile phone camera, to capture seed videos. 2. **Deep Learning Model**: The YOLOv8 neural network model is used for seed detection, combined with StrongSORT and ByteTrack object tracking algorithms for seed counting. 3. **Synthetic Data Generation**: Synthetic image datasets are generated through Domain Randomization (DR) to train the neural network model, overcoming the lack of real labeled data. ### Experimental Results - **Detection Accuracy**: On the test dataset, the model achieved detection accuracies of 93% and 92% for soybean and wheat seeds, respectively, with recall rates of 91% and 90%. - **Counting Accuracy**: Using the StrongSORT algorithm, the counting accuracies for soybean and wheat seeds were 95.2% and 93.2%, respectively; using the ByteTrack algorithm, the counting accuracies were 96.8% and 92.4%, respectively. ### Conclusion This study demonstrates the potential of using deep learning and synthetic image data generation techniques in the agricultural field, particularly for seed counting tasks. Through low-cost hardware and efficient algorithms, an affordable and accurate seed counting solution can be provided for small seed packaging enterprises. Future work will focus on improving the image capture device to reduce seed aggregation and occlusion issues and developing mobile applications suitable for different seed types.

Seed Kernel Counting using Domain Randomization and Object Tracking Neural Networks

Seed Phenotyping on Neural Networks using Domain Randomization and Transfer Learning

Robust soybean seed yield estimation using high-throughput ground robot videos

Apple Counting using Convolutional Neural Networks

Counting and Segmenting Sorghum Heads

DeepCorn: A semi-supervised deep learning method for high-throughput image-based corn kernel counting and yield estimation

TasselNetV2+: A Fast Implementation for High-Throughput Plant Counting From High-Resolution RGB Imagery

WheatNet: A lightweight convolutional neural network for high-throughput image-based wheat head detection and counting

Kiwifruit Counting Using Kiwidetector and Kiwitracker

Seed Classification using Synthetic Image Datasets Generated from Low-Altitude UAV Imagery

A Convolutional Neural Network-Based Method for Corn Stand Counting in the Field

Wheat Seed Detection and Counting Method Based on Improved YOLOv8 Model

Convolutional Neural Networks for Image-based Corn Kernel Detection and Counting

Rapeseed Flower Counting Method Based on GhP2-YOLO and StrongSORT Algorithm

Automatic kernel counting on maize ear using RGB images

Accurate machine learning-based germination detection, prediction and quality assessment of three grain crops

High-throughput soybean seeds phenotyping with convolutional neural networks and transfer learning

Segmentation and counting of wheat spike grains based on deep learning and textural feature

Segmentation and Phenotype Calculation of Rapeseed Pods Based on YOLO v8 and Mask R-Convolution Neural Networks

Real-time missing seedling counting in paddy fields based on lightweight network and tracking-by-detection algorithm

From Fields to Pixels: UAV Multispectral and Field-Captured RGB Imaging for High-Throughput Wheat Spike and Kernel Counting