No abstract

No abstract

Human facial emotion recognition (FER) has attracted the attention of the research community for its promising applications. Mapping different facial expressions to the respective emotional states are the main task in FER. The classical FER consists of two major steps: feature extraction and emotion recognition. Currently, the Deep Neural Networks, especially the Convolutional Neural Network (CNN), is widely used in FER by virtue of its inherent feature extraction mechanism from images. Several works have been reported on CNN with only a few layers to resolve FER problems. However, standard shallow CNNs with straightforward learning schemes have limited feature extraction capability to capture emotion information from high-resolution images. A notable drawback of the most existing methods is that they consider only the frontal images (i.e., ignore profile views for convenience), although the profile views taken from different angles are important for a practical FER system. For developing a highly accurate FER system, this study proposes a very Deep CNN (DCNN) modeling through Transfer Learning (TL) technique where a pre-trained DCNN model is adopted by replacing its dense upper layer(s) compatible with FER, and the model is fine-tuned with facial emotion data. A novel pipeline strategy is introduced, where the training of the dense layer(s) is followed by tuning each of the pre-trained DCNN blocks successively that has led to gradual improvement of the accuracy of FER to a higher level. The proposed FER system is verified on eight different pre-trained DCNN models (VGG-16, VGG-19, ResNet-18, ResNet-34, ResNet-50, ResNet-152, Inception-v3 and DenseNet-161) and well-known KDEF and JAFFE facial image datasets. FER is very challenging even for frontal views alone. FER on the KDEF dataset poses further challenges due to the diversity of images with different profile views together with frontal views. The proposed method achieved remarkable accuracy on both datasets with pre-trained models. On a 10-fold cross-validation way, the best achieved FER accuracies with DenseNet-161 on test sets of KDEF and JAFFE are 96.51% and 99.52%, respectively. The evaluation results reveal the superiority of the proposed FER system over the existing ones regarding emotion detection accuracy. Moreover, the achieved performance on the KDEF dataset with profile views is promising as it clearly demonstrates the required proficiency for real-life applications.

We investigated soil net nitrogen mineralization rate, above-and belowground biomass allocation, and nitrogen use in a Cryptomeria japonica plantation chronosequence. Total biomass accumulation showed an asymptotic accretion pattern, and the peak total biomass accumulation rate occurred approximately 30 years after afforestation. Soil net nitrogen mineralization rate was lowest 30 years after afforestation. Between years 30 and 88, net nitrogen mineralization increased again. These results indicate that an imbalance in soil nitrogen supply and plant nitrogen demand occurred approximately 30 years after afforestation. Furthermore, leaf nitrogen concentration, which was used as an index of plant nitrogen status, was lower in mature forest than in young forest, suggesting that mature stands did not take up nitrogen as successfully. If soil resources such as nitrogen limit plant growth, plants may increase biomass allocation to fine root structure; however, fine root biomass was not higher in 30-and 88-year-old stands than in younger stands, suggesting that changes in biomass allocation may not be effective against nitrogen deficiency in a C. japonica plantation chronosequence.