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.

Artificial LED source provides the possibility to regulate the lighting environment in plant factorys that use limited space to plant, aiming at high throughput and good quality. However, different parameters of light intensity, quality, and photoperiod will influence the growth and accumulation of bio-compounds in plants. In order to find the optimal setting of LED light for spinach planting, four group experiments were designed using the orthogonal testing method. According to the experimental results, for growth indexes including fresh weight, dry weight, root length and so on, photoperiod is the most influential factor, light intensity is the second, and light quality is the least. The best light mode (R:B = 4:1, photosynthetic photon flux density (PPFD) = 100 μmol∙m⁻²∙s⁻¹ and 13/11 h) among all eight possible combinations in the range was also determined. Furthermore, for quality indexes, including the soluble sugar content, protein content and so on, a new scoring method was introduced to make a comprehensive score for evaluating. Then, the light combination (R:B = 4:1, PPFD = 150 μmol∙m⁻²∙s⁻¹ and 9/15 h) in the range was found as the optimal scheme for spinach quality under those parameters. As there is trade-off between the optimal light parameters for growth and quality, it is necessary to achieve a balance between yield and quality of the plant during production. If farmers want to harvest spinach with larger leaf area and higher yield, they need to pay attention to the adjustment of the photoperiod and use a lower light intensity and a longer lighting time. If they do not mind the yield of the vegetable but want to improve the taste and nutrition of spinach products, they should pay more attention to the light intensity and use a higher light intensity and a shorter lighting time.

It is already known that the process of photosynthesis depends on the quality and intensity of light. However, the influence of the new light sources recently used in horticulture, known as Light Emitting Diodes (LEDs), on this process is not yet fully understood. Chlorophyll a fluorescence measurement has been widely used as a rapid, reliable, and noninvasive tool to study the efficiency of the photosystem II (PSII) and to evaluate plant responses to various environmental factors, including light intensity and quality. In this work, we tested the responses of the tomato photosynthetic apparatus to different light spectral qualities. Our results showed that the best performance of the photosynthetic apparatus was observed under a mixture of red and blue light (R7:B3) or a mixture of red, green and blue light (R3:G2:B5). This was demonstrated by the increase in the effective photochemical quantum yield of PSII (Y[II]), photochemical quenching (qP) and electron transport rate (ETR). On the other hand, the mixture of red and blue light with a high proportion of blue light led to an increase in non-photochemical quenching (NPQ). Our results can be used to improve the production of tomato plants under artificial light conditions. However, since we found that the responses of the photosynthetic apparatus of tomato plants to a particular light regime were cultivar-dependent and there was a weak correlation between the growth and photosynthetic parameters tested in this work, special attention should be paid in future research.

It is already known that there are many factors responsible for the successful grafting process in plants, including light intensity. However, the influence of the spectrum of light-emitting diodes (LEDs) on this process has almost never been tested. During the pre-grafting process tomato seedlings grew for 30 days under 100 μmol m^-2 s^-1 of mixed LEDs (red 70%+ blue 30%). During the post-grafting period, seedlings grew for 20 days under the same light intensity but the lightening source was either red LED, mixed LEDs (red 70% + blue 30%), blue LED or white fluorescent lamps. This was done to determine which light source(s) could better improve seedling quality and increase grafting success. Our results showed that application of red and blue light mixture (R7:B3) caused significant increase in total leaf area, dry weight (total, shoot and root), total chlorophyll/carotenoid ratio, soluble protein and sugar content. Moreover, this light treatment maintained better photosynthetic performance i.e. more effective quantum yield of PSII photochemistry Y(II), better photochemical quenching (qP), and higher electron transport rate (ETR). This can be partially explained by the observed upregulation of gene expression levels of PsaA and PsbA and the parallel protein expression levels. This in turn could lead to better functioning of the photosynthetic apparatus of tomato seedlings and then to faster production of photoassimilate ready to be translocated to various tissues and organs, including those most in need, i.e., involved in the formation of the graft union.

In recent years, much effort has been devoted to understanding the response of plants to various light sources, largely due to advances in industry light-emitting diodes (LEDs). In this study, the effect of different light modes on rocket (Eruca sativa. Mill.) photosynthetic performance and other physiological traits was evaluated using an orthogonal design based on a combination between light intensity, quality, and photoperiod factors. Some morphological and biochemical parameters and photosynthetic efficiency of the plants were analyzed. Plants grew in a closed chamber where three light intensities (160, 190, and 220 μmol m^-2 s^-1) provided by LEDs with a combination of different ratios of red, green, and blue (R:G:B- 7:0:3, 3:0:7, and 5:2:3) and three different photoperiods (light/dark -10/14 h, 12/12 h, and 14/10 h) were used and compared with white fluorescent light (control). This experimental setup allowed us to study the effect of 9 light modes (LM) compared to white light. The analyzes performed showed that the highest levels of chlorophyll a, chlorophyll b, and carotenoids occurred under LM4, LM3, and LM1, respectively. Chlorophyll a fluorescence measurement showed that the best effective quantum yield of PSII photochemistry Y(II), non-photochemical quenching (NPQ), photochemical quenching coefficient (qP), and electron transport ratio (ETR) were obtained under LM2. The data showed that the application of R7:G0:B3 light mode with a shorter photoperiod than 14/10 h (light/dark), regardless of the light intensity used, resulted in a significant increase in growth as well as higher photosynthetic capacity of rocket plants. Since, a clear correlation between the studied traits under the applied light modes was not found, more features should be studied in future experiments.

It is already known that there are many factors responsible for the successful formation of a graft union. However, the role of light has been little studied. In an anatomical study, Scanning Electronic Microscope (SEM) was used to explore the effects of different light-emitting diodes (LEDs) on graft union formation in grafted tomato. In addition, the expression genes related to Auxin hormone signaling pathway (SAUR67, AUX1, ARF30, and LAX3) was investigated. The obtained results showed that the concrescence process occurred faster under R7:B3 light conditions, as compared to blue (B) and white fluorescent (WFL) lights. Red light application caused a delay in the vascular tissue differentiation, which may lead to callus development on both sides, causing junctional failure and resulting in ineffective graft junctional arrangement. The expression of genes related to Auxin hormone significantly increased by R7:B3 application. We suggest that LED spectra affects the graft development of tomato plants and can improve the performance of grafted tomato seedlings.