ALBERT

Description: Intelligence has been considered as the major challenge in promoting economic potential and production efficiency of precision agriculture. In order to apply advanced deep-learning technology to complete various agricultural tasks in online and offline ways, a large number of crop vision datasets with domain-specific annotation are urgently needed. To encourage further progress in challenging realistic agricultural conditions, we present the CropDeep species classification and detection dataset, consisting of 31,147 images with over 49,000 annotated instances from 31 different classes. In contrast to existing vision datasets, images were collected with different cameras and equipment in greenhouses, captured in a wide variety of situations. It features visually similar species and periodic changes with more representative annotations, which have supported a stronger benchmark for deep-learning-based classification and detection. To further verify the application prospect, we provide extensive baseline experiments using state-of-the-art deep-learning classification and detection models. Results show that current deep-learning-based methods achieve well performance in classification accuracy over 99%. While current deep-learning methods achieve only 92% detection accuracy, illustrating the difficulty of the dataset and improvement room of state-of-the-art deep-learning models when applied to crops production and management. Specifically, we suggest that the YOLOv3 network has good potential application in agricultural detection tasks.

Description: Plants produce above- and below-ground biomass. However, our understanding of both production and decomposition of below-ground biomass is poor, largely because of the difficulties of accessing roots. Below-ground organic matter decomposition studies are scant and especially rare in the tropics. In this study, we used a litter bag experiment to quantify the mass loss and nutrient dynamics of decomposing twigs and small roots from an arbuscular mycorrhizal fungal associated tree, Parashorea chinensis Wang Hsie, in a tropical rain forest in Southwest China. Overall, twig litter decomposed 1.9 times faster than small roots (decay rate (k) twig = 0.255, root = 0.134). The difference in decomposition rates can be explained by a difference in phosphorus (P) concentration, availability, and use by decomposers or carbon quality. Twigs and small roots showed an increase in nitrogen concentration, with final concentrations still higher than initial levels. This suggests nitrogen transfer from the surrounding environment into decomposing twigs and small roots. Both carbon and nitrogen dynamics were significantly predicted by mass loss and showed a negative and positive relationship, respectively. Our study results imply that small roots carbon and nitrogen increase the resident time in the soil. Therefore, a better understanding of the carbon cycle requires a better understanding of the mechanisms governing below-ground biomass decomposition.

Description: Cultivated land productivity is a basic guarantee of food security. This study extracted the multiple cropping index (MCI) and most active days (MAD, i.e., days when the EVI exceeded a threshold) based on crop growth EVI curves to analyse the changes and potential characteristics of cultivated land productivity in Jiangsu Province during 2001–2017. The results are as follows: (1) The MCI of 83.8% of cultivated land remained unchanged in Jiangsu, the cultivated land with changed MCI (16.2%) was mainly concentrated in the southern and eastern coastal areas of Jiangsu, and the main cropping systems were single and double seasons. (2) The changes in cultivated land productivity were significant and had an obvious spatial distribution. The areas where the productivity of single cropping system changed occupied 67.8% of the total cultivated land of single cropping system, and the decreased areas (46.5%) were concentrated in southern Jiangsu. (3) For double cropping systems, the percentages of the changed productivity areas accounting for cultivated land were 82.7% and 73.3%. The decreased areas were distributed in central Jiangsu. In addition, the productivity of the first crop showed an overall (72%) increasing trend and increased areas (40.8%) of the second crop were found in northern Jiangsu. (4) During 2001–2017, cultivated land productivity greatly improved in Jiangsu. In the areas where productivity increased, the proportions of cultivated land with productivity potential space greater than 20% in single and double cropping systems were greater than 60% and 90%, respectively. In the areas where productivity decreased, greater than 25% and 75% of cultivated land had potential space in greater than 80% of the single and double cropping systems, respectively. This result shows that productivity still has much room for development in Jiangsu. This study provides new insight for studying cultivated land productivity and provides references for guiding agricultural production.

Description: The monitoring-blind area exists in the industrial park because of private interest and limited administrative power. As the atmospheric quality in the blind area impacts the environment management seriously, the prediction and inference of the blind area is explored in this paper. Firstly, the fusion network framework was designed for the solution of “Circumjacent Monitoring-Blind Area Inference”. In the fusion network, the nonlinear autoregressive network was set up for the time series prediction of circumjacent points, and the full connection layer was built for the nonlinear relation fitting of multiple points. Secondly, the physical structure and learning method was studied for the sub-elements in the fusion network. Thirdly, the spatio-temporal prediction algorithm was proposed based on the network for the blind area monitoring problem. Finally, the experiment was conducted with the practical monitoring data in an industrial park in Hebei Province, China. The results show that the solution is feasible for the blind area analysis in the view of spatial and temporal dimensions.