-
Categories
-
Pharmaceutical Intermediates
-
Active Pharmaceutical Ingredients
-
Food Additives
- Industrial Coatings
- Agrochemicals
- Dyes and Pigments
- Surfactant
- Flavors and Fragrances
- Chemical Reagents
- Catalyst and Auxiliary
- Natural Products
- Inorganic Chemistry
-
Organic Chemistry
-
Biochemical Engineering
- Analytical Chemistry
-
Cosmetic Ingredient
- Water Treatment Chemical
-
Pharmaceutical Intermediates
Promotion
ECHEMI Mall
Wholesale
Weekly Price
Exhibition
News
-
Trade Service
Application Case 1: Differentiation of Lycium barbarum with different fruit sizes
Application Case 2: Classification of Lycium barbarum from different origins based on hyperspectral images
Icotek Ecotech provides SpectraScan © non-destructive hyperspectral imaging detection solutions for food and drug testing and identification and other fields, and provides SpectrAPP ® spectral imaging technology innovation application project cooperation and technical services
.
Application Case 2: Classification of Lycium barbarum from different origins based on hyperspectral images
The study used a new hybrid convolutional neural network (New-Hybrid-CNN) architecture to make full use of pixel information to divide the wolfberry samples from each origin into 2500 3D patches, and randomly divide the wolfberry samples from different origins into 2,500 3D patches.
For the training set (70%) and the test set (30%), 20% of the training set is randomly selected as the validation set
.
Then use a homogeneous 3D convolution architecture with 3×3×3 convolution kernels to extract the spectral-spatial joint information from the hyperspectral cubic information, and then use depthwise separable convolution (DSC) to learn the spatial information, combined with 3D convolution And the advantages of DSC, the deep spectral-spatial joint information is effectively extracted, making the architecture more lightweight
.
Comparing 3D-CNN, HybridSN and SVM with the proposed method under the same parameters, it is found that New-Hybrid-CNN shows a better classification effect (as shown in the left figure below), and requires the least number of parameters and the shortest time.
and the most stable
.
For the training set (70%) and the test set (30%), 20% of the training set is randomly selected as the validation set
.
Then use a homogeneous 3D convolution architecture with 3×3×3 convolution kernels to extract the spectral-spatial joint information from the hyperspectral cubic information, and then use depthwise separable convolution (DSC) to learn the spatial information, combined with 3D convolution And the advantages of DSC, the deep spectral-spatial joint information is effectively extracted, making the architecture more lightweight
.
Comparing 3D-CNN, HybridSN and SVM with the proposed method under the same parameters, it is found that New-Hybrid-CNN shows a better classification effect (as shown in the left figure below), and requires the least number of parameters and the shortest time.
and the most stable
.
Icotek Ecotech provides SpectraScan © non-destructive hyperspectral imaging detection solutions for food and drug testing and identification and other fields, and provides SpectrAPP ® spectral imaging technology innovation application project cooperation and technical services
.
