A CNN-based lightweight ensemble model for detecting defective carrots

Xie, Weijun; Wei, Shuo; Zheng, Zhaohui; Yang, Dongya

doi:10.1016/j.biosystemseng.2021.06.008

Cited by 42 publications

(24 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The optimization algorithm, learning rate, number of epochs, batch size, loss function, and weight initialization are some crucial hyperparameters that need to be tuned before training of neural network to achieve better results. The improper selection of these hyperparameters may result in poor generalization performance (Xing et al, 2018), over tting, excess training time, over-consumption of computational resources (Xie et al, 2021), and may effects regularization signi cantly (Wilson and Martinez, 2003), etc.…”

Section: Training Of Cnn Modelsmentioning

confidence: 99%

“…Dropout is a regularization technique, used to avoid over tting in the neural networks by dropping out activation units randomly during training. This technique of regularization was successfully implemented in the fully connected layer of convolutional networks by various researchers for agricultural images (Khanramaki et al, 2021;Rahman et al, 2020;Waheed et al, 2020;Xie et al, 2021)…”

Section: Selection Of Dropout Ratementioning

confidence: 99%

“…A typical convolutional neural network consists of millions of trainable parameters (He et al, 2015;Ronneberger et al, 2015;Szegedy et al, 2015) out of which possibly some parameters may contribute meager or negligible to the output of the network and hence, can be removed without affecting the e ciency of the network (Molchanov et al, 2017). To remove the non-performing parameters of a CNN, ablation experiments were conducted by multiple researchers (Adhikari et al, 2019;Bao et al, 2021;Xie et al, 2021;You et al, 2020) in which certain components/layers of the network were pruned. The goal of the ablation study is to reduce the network size, computational expensiveness, training time, and inference time, providing the actual e ciency of the network is retained or degraded insigni cantly.…”

Section: Ablation Study Of Architecturesmentioning

confidence: 99%

“…A convolutional neural network is such a technique that can extract high-dimensional features automatically through self-learned features. Using this convolutional neural network technique, numerous researchers achieved promising results for classi cation (Dyrmann et al, 2016;Jiang et al, 2019;Khanramaki et al, 2021;Waheed et al, 2020;Xie et al, 2021;Zhang et al, 2019) as well as for semantic segmentation (Azizi et al, 2020;Chen et al, 2020;Kang et al, 2021;Kestur et al, 2019;Tassis et al, 2021;Zabawa et al, 2020;Zou et al, 2021) tasks. For classi cation tasks, at the end of convolutional layers, dense layers are added while for semantic segmentation, fully convolutional neural networks are used in which a speci c class is assigned to each pixel based on high features learned during the training of convolutional neural network models.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Lightweight convolutional neural network models for semantic segmentation of in-field cotton bolls

Singh

Tewari

Biswas

et al. 2022

Preprint

View full text Add to dashboard Cite

Introduction Robotic harvesting of cotton bolls will incorporate the benefits of manual picking as well as mechanical harvesting. For robotic harvesting, in-field cotton segmentation with minimal errors is desirable which is a challenging task. In the present study, three lightweight fully convolutional neural network models were developed for the semantic segmentation of in-field cotton bolls. Materials and Methods Model 1 does not include any residual or skip connections, while model 2 consists of residual connections to tackle the vanishing gradient problem and skip connections for feature concatenation. Model 3 along with residual and skip connections, consists of filters of multiple sizes. Effects of filter size and the dropout rate were studied. Results All proposed models segment the cotton bolls successfully with the cotton IoU (intersection-over-union) value of above 88%. The highest cotton IoU of 91.03% was achieved by model 2. The proposed models achieved F1-score and pixel accuracy values greater than 95% and 98%, respectively. The developed models were compared with existing state-of-the-art networks namely VGG19, ResNet18, and InceptionV3. Despite having a limited number of trainable parameters, the proposed models achieved mIoU (mean intersection-over-union) of 93.84%, 94.15%, and 94.65% against the mIoU values of 95.39%, 96.54%, and 96.37% obtained using state-of-the-art networks. The segmentation time for the developed models was reduced up to 52% compared to state-of-the-art networks. Conclusion The developed lightweight models segmented the in-field cotton bolls comparatively faster and with greater accuracy. Hence, developed models can be deployed to cotton harvesting robots for real-time recognition of in-field cotton bolls for harvesting.

show abstract

Section: Training Of Cnn Modelsmentioning

confidence: 99%

Section: Selection Of Dropout Ratementioning

confidence: 99%

Section: Ablation Study Of Architecturesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Lightweight convolutional neural network models for semantic segmentation of in-field cotton bolls

Singh

Tewari

Biswas

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…It has achieved good results in tomato defect detection ( da Costa et al, 2020 ), apple defect detection ( Zhang et al, 2017 ), and litchi surface micro-damage detection ( Wang et al, 2016 ). The machine vision algorithm combined with the deep learning model has predominant robustness in carrot defect detection ( Xie et al, 2021 ). Choosing the appropriate learning algorithm for a specific problem is crucial for vegetable defect detection based on the deep learning algorithm.…”

Section: Introductionmentioning

confidence: 99%

Surface Defect Detection of Cabbage Based on Curvature Features of 3D Point Cloud

Zhang

Yin

et al. 2022

Front. Plant Sci.

View full text Add to dashboard Cite

The dents and cracks of cabbage caused by mechanical damage during transportation have a direct impact on both commercial value and storage time. In this study, a method for surface defect detection of cabbage is proposed based on the curvature feature of the 3D point cloud. First, the red-green-blue (RGB) images and depth images are collected using a RealSense-D455 depth camera for 3D point cloud reconstruction. Then, the region of interest (ROI) is extracted by statistical filtering and Euclidean clustering segmentation algorithm, and the 3D point cloud of cabbage is segmented from background noise. Then, the curvature features of the 3D point cloud are calculated using the estimated normal vector based on the least square plane fitting method. Finally, the curvature threshold is determined according to the curvature characteristic parameters, and the surface defect type and area can be detected. The flat-headed cabbage and round-headed cabbage are selected to test the surface damage of dents and cracks. The test results show that the average detection accuracy of this proposed method is 96.25%, in which, the average detection accuracy of dents is 93.3% and the average detection accuracy of cracks is 96.67%, suggesting high detection accuracy and good adaptability for various cabbages. This study provides important technical support for automatic and non-destructive detection of cabbage surface defects.

show abstract

Visualization and location estimation of defective parts of industrial products using convolutional autoencoder

Arima

Nagata

Shimizu

et al. 2022

Artif Life Robotics

View full text Add to dashboard Cite

A CNN-based lightweight ensemble model for detecting defective carrots

Cited by 42 publications

References 30 publications

Lightweight convolutional neural network models for semantic segmentation of in-field cotton bolls

Lightweight convolutional neural network models for semantic segmentation of in-field cotton bolls

Surface Defect Detection of Cabbage Based on Curvature Features of 3D Point Cloud

Visualization and location estimation of defective parts of industrial products using convolutional autoencoder

Contact Info

Product

Resources

About