China's soybean supply and demand are seriously imbalanced. It is crucial to improve the level of soybean breeding. Hundred-grain weight is one of the most essential phenotypic parameters for crop breeding. Accurate soybean seed counting is a key step for 100-grain weight. There are several seed counting methods, which have their own limitations one way or the other. Among these, manual counting is time-consuming, electronic automatic seed counter devices are expensive and their counting speed is very slow, and the traditional digital image processing techniques are not suitable for seed counting based on individual pod images. This paper attempted to develop a method that would combine the density estimation-based methods and the convolution neural network (CNN)-based methods to accurately estimate the seed count from an individual soybean pod image with a single perspective. In this paper, we first introduced a new large-scale seed counting dataset, named Soybean-pod. The dataset contains 500 annotated pod images with a total of 32 126 seeds and is the largest annotated dataset for soybean seed counting so far. Simultaneously, we used annotation information to generate a ground-truth density map by convolving a Gaussian kernel and, then, devised a simple but effective method that would elucidate pod images to a seed density map using a two-column CNN (TCNN) and thus accomplish seed counting ultimately. We conducted relevant experiments from three aspects on the new dataset to verify the effectiveness of our model and method, which provided 13.21 mean absolute error (MAE) and 17.62 mean squared error (mse). In addition, our research results showed that deep learning techniques can be easily adapted to precision tasks for plant phenotyping and breeding purposes.

Effective and efficient fruit detection is considered crucial for designing automated robot (AuRo) for yield estimation, disease control, harvesting, sorting, and grading. Several fruit detection schemes for designing AuRo have been developed during the last decades. However, conventional fruit detection methods are deficient in the real-time response, accuracy, and extensibility. This paper proposes an improved multi-task cascaded convolutional network-based intelligent fruit detection method. This method has the capability to make the AuRo work in real time with high accuracy. Moreover, based on the relationship between the diversity samples of the dataset and the parameters of neural networks' evolution, this paper presents an improved augmented method, a procedure that is based on image fusion to improve the detector performance. The experiment results demonstrated that the proposed detector performed immaculately both in terms of accuracy and time-cost. Furthermore, the extensive experiment also demonstrated that the proposed technique has the capacity and good portability to work with other akin objects conveniently.

Generated topic label, an alternative representation of topics learned by topic model, is widely used to help the user interpret the topics more efficiently. A major challenge now is to label a discovered topic accurately in an objective way. This article introduces a novel graph-based ranking model (TLRank), to find a meaningful topic label with high Relevance, Coverage, and Discrimination. The model applies a specific strategy that suppresses or enhances the matrix transition probability according to the textual similarity between vertices (sentences) and the characteristics of vertices respectively. Moreover, to boost diversity and enhance performance, TLRank scores the candidate sentences and refrains redundancy of topic labels simultaneously in a single labeling process. In our experiments, the evaluation results showed that the TLRank model significantly and consistently outperformed the prevailing state-of-the-art and classic models in topic labeling task.

Fuji apples are one of the most important and popular economic crops worldwide in the fruit industry. Nowadays, there is a huge imbalance between the urgent demand of precise automated sorting models of fruit ripeness grades due to the increasing consumption levels and the limitations of most existing methods. In this regard, this paper presents a novel CNN-based fine-grained lightweight architecture for the task of Fuji apple maturity classification (FGAL-MC). Our proposed FGAL-MC architecture has three advantages compared with related previous research works. Firstly, we established a novel Fuji apple maturity dataset. We investigated the Fuji apple’s different growth stages using image samples that were captured in open-world orchard environments, which have the benefit of being able to guide the related methods to be more suitable for the practical working environment. Secondly, because maturity grades are difficult to discriminate due to the issues of subtle expression differences, as well as the various challenging disadvantages for the unstructured surroundings, we designed our network as a fine-grained classification architecture by introducing an attention mechanism to learn class-specific regions and discrimination. Thirdly, because the number of parameters of an architecture determines the time-cost and hardware configuration to some extent, we designed our proposed architecture as a lightweight structure, which is able to be applied or promoted for actual agriculture field operations. Finally, comprehensive qualitative and quantitative experiments demonstrated that our presented method can achieve competitive results in terms of accuracy, precision, recall, F1-score, and time-cost. In addition, extensive experiments indicated our proposed method also has outstanding performance in terms of generalization ability.

Pedestrian detection is a key problem for automatic driving, and results have been improved significantly via deep convolutional networks. However, there is still room to improve the performance of pedestrian detection by carefully dealing with some critical issues. To take advantages of more discriminative information for pedestrian detection, we propose a novel architecture to auto-choose semantic as well as specific information among the feature maps at different levels and integrate valuable information among the feature maps in multi-scales. Particularly, our architecture consists of feature maps concatenating in different levels and feature maps integrating with multi-scales. Both the operations are equipped with a competitive attention block. The architecture has the ability to obtain more efficient and discriminate features for pedestrian detection. In comparison with the other prevailing models, our architecture provides superior performance. The promising results achieved through experimentation with this architecture achieve a new state-of-the-art on Caltech dataset.

Depth estimation is one of the bottleneck parts for harvesting robots to determine whether the operation of grasping or picking succeeds or not directly. This paper proposed a novel disparity completion method combined with bilateral filtering and pyramid fusion to improve the issues of incorrect outputs due to the missed or wrong matching when achieving 3D position from 2D images in open-world environments. Briefly, our proposed method has two significant advantages in general. Firstly, occlusion between leaves, branches, and fruits is a universal phenomenon in unstructured orchard environments, which results in the most depth estimation algorithms facing great challenges to obtain accurate outputs in these occluded regions. To alleviate these issues, unlike other research efforts that already exist, we optimized the semi-global matching algorithm to obtain high accuracy sparse values as an initial disparity map; then, an improved bilateral filtering algorithm is proposed to eliminate holes and discontinuous regions caused by occlusion to obtain precise and density disparity outputs. Secondly, due to taking the practical high-efficiency requirements of the automated harvesting robot in its working status into consideration, we attempted to merge multiple low-resolution bilateral filtering results through the pyramid fusion model which goes beyond the trade-off mechanism to improve the performance of both accuracy and time cost. Finally, a prototype harvesting robot was designed to conduct experiments at three kinds of different distances (0.6~0.75 m, 1~1.2 m, and 1.6~1.9 m). Experiment results showed that our proposed method achieved density disparity maps and eliminated holes and discontinuous defects in the disparity map effectively. The average absolute error of our proposed method is 3.2 mm, and the average relative error is 1.79%. In addition, the time cost is greatly reduced more than 90%. Comprehensive experimental results demonstrate that our proposed algorithm provides a potential possibility for designing harvesting.