Skip to main content

Solution Design

Programme

1. Data Preprocessing​

Data Preprocessing

2. Model Design​

Model design is implemented based on Keras components, including Keras' public property packages layers and inputs for easy construction of your own network models; classic network model structures such as VGG16, MobileNet, Inception are encapsulated; custom-designed models according to specified formats.

Model Design

3. Algorithm Library​

Numerical algorithms: classification algorithms, regression algorithms, clustering algorithms; visual algorithms: image classification, object recognition, image segmentation, face recognition, keypoint detection; simulation algorithms: cloud chart regression algorithm.

Algorithm Library

4. Model Library​

Model Library

For trained models, the model library module will generate corresponding cards based on the model name, and create a version component area for the model library according to the order in which the models were created. The model name and version in the attribute area correspond to those in the model library.

Completed Model

Model Prediction Module Model Prediction

5. Solution Instances​

Each run of the solution is recorded in the solution instance, which includes instance details, instance logs, and instance data. Instance details contain basic information about the instance run, model information, and resource information; instance logs are generated during the operation of the instance; instance data is generated during the process of the instance run.

Solution Instance