LFPNet: Lightweight network on real point sets for fruit classification and segmentation

Yu, Qirui; Yang, Huijun; Gao, Yangbo; Ma, Xinrui; Chen, Guochao; Wang, Xin

doi:10.1016/j.compag.2022.106691

Cited by 13 publications

(5 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Such problematic class distribution could severely affect the accuracy of the trained model. Yu et al's work [21] is the most similar work to our method. However, their work does not discuss the multi-sensor data fusion and training under imbalance class, which are two critical problems of 3D semantic segmentation orchard scenes.…”

Section: Introductionmentioning

confidence: 80%

“…The experiment results suggested the best segmentation accuracy of 94%, and the mIoU can reach 74%. Yu et al designed LFPNet that can directly consume fruit point clouds in real scenes to deal with classification error, incomplete segmentation, and low-efficiency [21]. The final results achieved an average segmentation accuracy and mIOU of 80.2% and 76.4%, respectively.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Semantic Segmentation of Fruits on Multi-sensor Fused Data in Natural Orchards

Kang¹,

Wang²

2022

Preprint

View full text Add to dashboard Cite

Semantic segmentation is a fundamental task for agricultural robots to understand the surrounding environments in natural orchards. The recent development of the LiDAR techniques enables the robot to acquire accurate range measurements of the view in the unstructured orchards. Compared to RGB images, 3D point clouds have geometrical properties. By combining the LiDAR and camera, rich information on geometries and textures can be obtained. In this work, we propose a deep-learning-based segmentation method to perform accurate semantic segmentation on fused data from a LiDAR-Camera visual sensor. Two critical problems are explored and solved in this work. The first one is how to efficiently fused the texture and geometrical features from multi-sensor data. The second one is how to efficiently train the 3D segmentation network under severely imbalance class conditions. Moreover, an implementation of 3D segmentation in orchards including LiDAR-Camera data fusion, data collection and labelling, network training, and model inference is introduced in detail. In the experiment, we comprehensively analyze the network setup when dealing with highly unstructured and noisy point clouds acquired from an apple orchard. Overall, our proposed method achieves 86.2% mIoU on the segmentation of fruits on the high-resolution point cloud (100k-200k points). The experiment results show that the proposed method can perform accurate segmentation in real orchard environments.

show abstract

Section: Introductionmentioning

confidence: 80%

Section: Introductionmentioning

confidence: 99%

Semantic Segmentation of Fruits on Multi-sensor Fused Data in Natural Orchards

Kang¹,

Wang²

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…In reference [20], convolutional neural network was used to detect and classify leaf diseases, the classification accuracy was relatively high. There are also many deep learning networks applied in fruit classification, such as [21][22][23][24][25][26][27], and achieved good results. Alharbi et al [28] adopted feature fusion deep learning to carry out automatic fruit classification and achieved good results.…”

Section: A Fruit Image Classification Methodsmentioning

confidence: 99%

Early Detection of Ripeness for the Picking of Xanthoceras Sorbifolium Using Feature Excitation-Based Broad Learning System

Dan,

Zuchen,

Liying

et al. 2024

IEEE Access

View full text Add to dashboard Cite

The seed oil of Xanthoceras sorbifolium is a new kind of vegetable oil which is beneficial to the body. However, during the ripening process, if not picked properly, resulting in seed waste and economic loss. Therefore, selecting the right picking time is of great significance for improving seed yield, reducing waste of labor and capital costs, and improving economic benefits. It is very challenging to achieve real-time and accurate classification of the images in the ripening stage of Xanthoceras sorbifolium with similar color, different shapes and serious background interference. So as to extract effective fruit features and improve the classification efficiency, a broad learning image classification method based on feature excitation (BL-SENet) was proposed in this paper. Firstly, a broad learning system (BLS) was constructed to extract the fruit features based on node activation function for the input layer. Secondly, feature excitation is carried out based on SENet, and the learning weights of features extracted based on broad learning mechanism are re-calibrated to improve the accuracy of network classification. Finally, based on the feature calibration, image classification is carried out by taking full advantage of the fast broad learning system. It is tested through experiments, the training accuracy of the proposed method is 100%, and the test accuracy is more than 80%, and it is the fastest among the comparison methods (except BLS). In order to promote the development of intelligent agriculture and realize intelligent mechanical picking, it provides effective visual information.

show abstract

“…In recent years, deep learning models have achieved great success in many computer vision problems (Ibrahim et al, 2022; Yu et al, 2022). Convolutional neural network (CNN) is a deep learning algorithm that uses multiple layers of linear information processing to extract high‐level features and becomes an end‐to‐end solution (Lecun et al, 2015).…”

Section: Introductionmentioning

confidence: 99%

Automated tea quality identification based on deep convolutional neural networks and transfer learning

Zhang

Wang

et al. 2023

J Food Process Engineering

View full text Add to dashboard Cite

Different quality grades of tea tend to have a high degree of similarity in appearance. Traditional image‐based identification methods have limited effects, while complex deep learning architectures require much data and long‐term training. In this paper, two tea quality identification methods based on deep convolutional neural networks and transfer learning are proposed. Different types and quality of tea images are collected by a self‐designed computer vision system to form a data set, which is small‐scale and of high inter‐ and intraclass similarity. The first method uses three simplified convolutional neural network (CNN) models with different image input sizes to identify the quality of tea. The second method performs transfer learning to identify the tea quality by fine‐tuning the mature AlexNet and ResNet50 architecture. Classification performance and model complexity are measured and compared. The related application software is also developed. The results show that the performance of the CNN models and the transfer learning models are close, and both can achieve high identification accuracy. However, the complexity of the CNN models is two to three orders of magnitude lower than that of the transfer learning models. The study shows that deep CNNs and transfer learning have great potential to be rapid and effective methods for automated tea quality identification tasks with high inter‐ and intrasimilarity.

show abstract

LFPNet: Lightweight network on real point sets for fruit classification and segmentation

Cited by 13 publications

References 29 publications

Semantic Segmentation of Fruits on Multi-sensor Fused Data in Natural Orchards

Semantic Segmentation of Fruits on Multi-sensor Fused Data in Natural Orchards

Early Detection of Ripeness for the Picking of Xanthoceras Sorbifolium Using Feature Excitation-Based Broad Learning System

Automated tea quality identification based on deep convolutional neural networks and transfer learning

Contact Info

Product

Resources

About