• Korean Journal of Soil Science and Fertilizer
• /
• v.40 no.5
• /
• pp.383-391
• /
• 2007

This study was conducted for an assessment through the estimation of soil loss by each catchment classified by soil catena. Ten catchments, which are Geumgang21, Namgang03, Dongjincheon, Gapyongcheon01, Gyongancheon02, Geumgang16, Byongsungcheon01, Daesincheon, Bukcheon02, Youngsangang08, were selected from the hydrologic unit map and the detailed soil digital map (1:25,000) for this study. The catchments like Geumgang21, Namgang03, Dongjincheon, Gapyongcheon01 and Gyongancheon02 were mainly composed with soils originated from gneiss. The catchments like Geumgang16, Byongsungcheon01, Daesincheon, Bukcheon02 and Youngsangang08 were mainly composed with soils originated from granites. The grades, which are divided into seven grades with A(very tolerable), B(tolerable), C(moderate), D(low), E(high), F(severe), G(very severe), of soil erosion estimated by USLE in catchments were distributed in most A and B because of paddy land and forestry. In detailed, the soil erosion grade of catchments mainly distributing soils originated from gneiss showed more the distribution of B and C than it of catchments mainly distributing soils originated from granites. The reason of results would be derived from topographic characteristics of soils originated from gneiss located at mountainous. The soil loss according to soil catena linked with Songsan and Jigok series, which are soils originated from gneiss was calculated with $7.66ton\;ha^{-1}\;yr^{-1}$. The soil loss of Geumgang16, Byongsungcheon01, Daesincheon, Bukcheon02 which have the soil catena linked with Samgak and Sangju soil series originated from granite, was calculated with $5.55ton\;ha^{-1}\;yr^{-1}$. The soil loss of Youngsangang08 which have the soil catena linked with Songjung and Baeksan soil series originated from granite was calculated with $9.6ton\;ha^{-1}\;yr^{-1}$, but the conclusion on soil loss in this kind of soil catena would be drawn from the analysis of more catchments. In conclusion, the results of this study inform that the classification of soil catena by catchments and estimation of soil loss according to soil catena would be effective for analysis on the grade of non-point pollution by soil erosion in a catchment.

• Journal of Intelligence and Information Systems
• /
• v.24 no.1
• /
• pp.205-225
• /
• 2018

Convolutional Neural Network (ConvNet) is one class of the powerful Deep Neural Network that can analyze and learn hierarchies of visual features. Originally, first neural network (Neocognitron) was introduced in the 80s. At that time, the neural network was not broadly used in both industry and academic field by cause of large-scale dataset shortage and low computational power. However, after a few decades later in 2012, Krizhevsky made a breakthrough on ILSVRC-12 visual recognition competition using Convolutional Neural Network. That breakthrough revived people interest in the neural network. The success of Convolutional Neural Network is achieved with two main factors. First of them is the emergence of advanced hardware (GPUs) for sufficient parallel computation. Second is the availability of large-scale datasets such as ImageNet (ILSVRC) dataset for training. Unfortunately, many new domains are bottlenecked by these factors. For most domains, it is difficult and requires lots of effort to gather large-scale dataset to train a ConvNet. Moreover, even if we have a large-scale dataset, training ConvNet from scratch is required expensive resource and time-consuming. These two obstacles can be solved by using transfer learning. Transfer learning is a method for transferring the knowledge from a source domain to new domain. There are two major Transfer learning cases. First one is ConvNet as fixed feature extractor, and the second one is Fine-tune the ConvNet on a new dataset. In the first case, using pre-trained ConvNet (such as on ImageNet) to compute feed-forward activations of the image into the ConvNet and extract activation features from specific layers. In the second case, replacing and retraining the ConvNet classifier on the new dataset, then fine-tune the weights of the pre-trained network with the backpropagation. In this paper, we focus on using multiple ConvNet layers as a fixed feature extractor only. However, applying features with high dimensional complexity that is directly extracted from multiple ConvNet layers is still a challenging problem. We observe that features extracted from multiple ConvNet layers address the different characteristics of the image which means better representation could be obtained by finding the optimal combination of multiple ConvNet layers. Based on that observation, we propose to employ multiple ConvNet layer representations for transfer learning instead of a single ConvNet layer representation. Overall, our primary pipeline has three steps. Firstly, images from target task are given as input to ConvNet, then that image will be feed-forwarded into pre-trained AlexNet, and the activation features from three fully connected convolutional layers are extracted. Secondly, activation features of three ConvNet layers are concatenated to obtain multiple ConvNet layers representation because it will gain more information about an image. When three fully connected layer features concatenated, the occurring image representation would have 9192 (4096+4096+1000) dimension features. However, features extracted from multiple ConvNet layers are redundant and noisy since they are extracted from the same ConvNet. Thus, a third step, we will use Principal Component Analysis (PCA) to select salient features before the training phase. When salient features are obtained, the classifier can classify image more accurately, and the performance of transfer learning can be improved. To evaluate proposed method, experiments are conducted in three standard datasets (Caltech-256, VOC07, and SUN397) to compare multiple ConvNet layer representations against single ConvNet layer representation by using PCA for feature selection and dimension reduction. Our experiments demonstrated the importance of feature selection for multiple ConvNet layer representation. Moreover, our proposed approach achieved 75.6% accuracy compared to 73.9% accuracy achieved by FC7 layer on the Caltech-256 dataset, 73.1% accuracy compared to 69.2% accuracy achieved by FC8 layer on the VOC07 dataset, 52.2% accuracy compared to 48.7% accuracy achieved by FC7 layer on the SUN397 dataset. We also showed that our proposed approach achieved superior performance, 2.8%, 2.1% and 3.1% accuracy improvement on Caltech-256, VOC07, and SUN397 dataset respectively compare to existing work.