Applications of deep-learning approaches in horticultural research: a review

Abstract

Deep learning is known as a promising multifunctional tool for processing images and other big data. By assimilating large amounts of heterogeneous data, deep-learning technology provides reliable prediction results for complex and uncertain phenomena. Recently, it has been increasingly used by horticultural researchers to make sense of the large datasets produced during planting and postharvest processes. In this paper, we provided a brief introduction to deep-learning approaches and reviewed 71 recent research works in which deep-learning technologies were applied in the horticultural domain for variety recognition, yield estimation, quality detection, stress phenotyping detection, growth monitoring, and other tasks. We described in detail the application scenarios reported in the relevant literature, along with the applied models and frameworks, the used data, and the overall performance results. Finally, we discussed the current challenges and future trends of deep learning in horticultural research. The aim of this review is to assist researchers and provide guidance for them to fully understand the strengths and possible weaknesses when applying deep learning in horticultural sectors. We also hope that this review will encourage researchers to explore some significant examples of deep learning in horticultural science and will promote the advancement of intelligent horticulture.

Fig. 1. A DCNN architecture.

The model contains one input layer, four convolutional layers, four ReLU components, two stochastic pooling layers, two fully connected layers and one softmax regression output layer. Source: ref.

Fig. 2. A DCNN framework.

The performance and stability are improved by the batch normalization layer. Overfitting is prevented by the dropout layer. Global average pooling can adapt to different input image sizes. Source: ref.

Fig. 3. VGG-16 model for image recognition.

a The input images. b Visualization of the feature extraction results after each convolution (conv), pooling (pool) or fully connected (fc) layer. c The top-k prediction results. Source: ref.

Fig. 4. Structure of an RNN.

The information of the RNN propagates upwards from the initial input state. The only feedback of data is from the output neurons to the hidden neurons. The activation functions for the hidden and output neurons are the hyperbolic tangent and pure linear functions, respectively. Source: ref.

Fig. 5. The SAE-FNN architecture.

a The autoencoder structure, b SAE is pretrained in an unsupervised manner with random pixel spectra, c SAE-FNN is fine-tuned in a supervised manner with mean spectra and firmness (or SSC). Source: ref.

Fig. 6. The Mask R-CNN architecture.

The architecture consists of two parts: the backbone and the network head. Source: ref.