Product Features
Model Testing
Optimized for Edge AI
Reporting
No Coding
Labelling Tool
Training Engine
Advanced Framework
Model Testing
With built-in testing interface, ODHUB allows users to test the model performance before final deployment. Users can select any built-in video capture devices in their system and test how the trained model performs detection tasks in real-time.
Design AI model
TEST
TEST TRAINED NEURAL NETWORK MODEL FOR DEPLOYMENNT
LOAD DATASET
LOAD IMAGES INTO ODHUB
EVALUATE
EVALUATE THE TRAINED NEURAL NETWORK
TRAIN
TRAIN A NEURAL NETWORK
CONFIGURE
CONFIGURE A NEURAL NETWORK STRUCTURE
EVALUATE
After the training process, user can move to the evaluation page to compare different experiments. ODHUB provides an intuitive report that helps user to choose the best model for his/her object detection application. This information includes model accuracy, speed, and performance.
TRAIN
ODHUB will automatically train model using GPU if available. During the training and test process, both training loss, evaluation loss, and the mAP information are reported to the user. User can also choose another experiment with a different type of deep learning model to train.
CONFIGURE
User only needs to configure a particular experiment with correct training parameters such as model type, number of steps, batch size, learning rate. ODHUB will automatically set other parameters that are necessary for the training process.
TEST
ODHUB provides the built-in testing interface that allow users to test the model performance before final deployment. User can choose to host the model on ANSCloud and perform model inference via web service API calls, or download the model to deploy on local hardware.
LOAD DATASET
The data folder structure includes images file and XML file (contains label information). ODHUB will automatically validate the accuracy and sufficiency of the data. ODHUB also provides a built-in labeling tool, which allow users to perform the labeling task if necessary
EVALUATE
After the training process, user can move to the evaluation page to compare different experiments. ODHUB provides an intuitive report that helps user to choose the best model for his/her object detection application. This information includes model accuracy, speed, and performance.
TRAIN
ODHUB will automatically train model using GPU if available. During the training and test process, both training loss, evaluation loss, and the mAP information are reported to the user. User can also choose another experiment with a different type of deep learning model to train.
CONFIGURE
User only needs to configure a particular experiment with correct training parameters such as model type, number of steps, batch size, learning rate. ODHUB will automatically set other parameters that are necessary for the training process.
TEST
ODHUB provides the built-in testing interface that allow users to test the model performance before final deployment. User can choose to host the model on ANSCloud and perform model inference via web service API calls, or download the model to deploy on local hardware.
LOAD DATASET
The data folder structure includes images file and XML file (contains label information). ODHUB will automatically validate the accuracy and sufficiency of the data. ODHUB also provides a built-in labeling tool, which allow users to perform the labeling task if necessary
EVALUATE
After the training process, user can move to the evaluation page to compare different experiments. ODHUB provides an intuitive report that helps user to choose the best model for his/her object detection application. This information includes model accuracy, speed, and performance.