Add microclimate vision case

cases/microclimate-vision/index.qmd

@@ -0,0 +1,14 @@
+---
+title: Multimodal prediction of climatic parameters using street-level and satellite imagery
+author:
+    - Kunihiko Fujiwara et al.
+description: |
+   High-resolution microclimate data is essential for capturing spatio-temporal heterogeneity of urban climate and heat health management. However, previous studies have relied on dense measurements that require significant costs for equipment, or on physical simulations demanding intensive computational loads. As a potential alternative to these methods, we propose a multimodal deep learning model to predict microclimate at a high spatial and temporal resolution based on street-level and satellite imagery. This model consists of LSTM and ResNet-18 architectures, and predicts air temperature, relative humidity, wind speed, and global horizontal irradiance. The original article can be read online <a href="https://doi.org/10.1016/j.scs.2024.105733">here</a>.
+date: 2024-08-20
+image: banner.jpg
+categories: ["Deep Learning"]
+title-block-banner: false
+toc: false
+---
+
+{{< pdf /resources/fujiwara2024microclimate.pdf width=100% height=800 >}}

cases/rice-grain-classification/index.qmd

@@ -1,7 +1,7 @@
 ---
 title: Classifying rice grains using deep learning
 author:
-    - Farshad Farahnakian et al.</a>
+    - Farshad Farahnakian et al.
 description: |
    Rice, one of the world’s most significant agricultural products, is crucial for human nutrition, economies, and various industrial sectors. Classifying rice varieties, an essential part of rice supply management is often time-consuming, energy-intensive, and expensive. With over 120,000 rice varieties categorized by the International Rice Research Institute based on milling degree, kernel size, starch content, and flavour, the need for automation in rice grain classification is evident. We evaluated the performance of various contemporary deep-learning models, including Residual Network (ResNet), Visual Geometry Group (VGG) network, EfficientNet, and MobileNet. These models were tested on a dataset comprising 75,000 images, classified into five different rice categories. We assessed each model using established evaluation metrics such as accuracy, F1 score, precision, recall, and per-class accuracy. Our findings showed that the EfficientNet-based model delivered the highest accuracy (99.67%), while the MobileNet-based model excelled in the speed of classification (2556 s). We concluded that, compared to traditional machine learning methods, the models employed in our study are highly scalable and capable of managing large volumes of complex data with millions of features and samples. The original article can be read online <a href="https://doi.org/10.1016/j.jafr.2023.100890">here</a>.
 date: 2023-11-28