English | 简体中文

Welcome to SSCMA Model Zoo. We provide a series of pre-trained models for different application scenarios for you to use, with SSCMA, you can test or inference on these models and easily deploy them to edge computing devices.

SSCMA Model Zoo focuses on providing models trained on SSCMA optimized neural networks, which are tailored to real-world application scenarios and enable faster and more accurate inference on embedded devices.

Currently, SSCMA Model Zoo provides pre-trained models for the following application scenarios:

Model	Colab
Gender_Detection_Swift-YOLO_192
Digital_Meter_Water_Swift-YOLO_192
Apple_Detection_Swift-YOLO_192
Pet_Detection_Swift-YOLO_192
person_Detection_Swift-YOLO_Nano_192
person_Detection_Swift-YOLO_192
Face_Detection_Swift-YOLO_96
COCO_Detection_Swift-YOLO_320
Gesture_Detection_Swift-YOLO_192
Digital_Meter_Electricity_Swift-YOLO_192
Strawberry_Detection_Swift-YOLO_192

Model	Colab
MNIST_Classification_MobileNetV2_0.5_Rep_32
Gender_Classification_MobileNetV2_0.35_Rep_64
Person_Classification_MobileNetV2_0.35_Rep_64
Person_Classification_MobileNetV2_0.35_Rep_96
Person_Classification_MobileNetV2_0.35_Rep_32
CIFAR-10_Classification_MobileNetV2_0.35_Rep_32

If you need any pre-trained model for a specific dataset in a specific scenario, feel free to submit a feature request to Issues.

If you wish to use the model provided by SSCMA Model Zoo, we recommend that you follow the following steps:

1. Based on actual needs, select corresponding application scenario and choose appropriate neural networks. You can browse the test results that we provide for decision.
2. Download the selected pre-trained model. For public pre-trained models, you can directly download them through the model link in the test results table.
3. Please refer to SSCMA Documentation - Deployment Example to deploy on edge computing devices. You can also use SSCMA to run our models on your computer, reproduce our test results or infer directly.

If you encounter any problem with pre-trained models in SSCMA Model Zoo, we recommend that you first follow these steps to troubleshoot.

1. Check the correctness of the downloaded model. The end of the pre-trained model file name contains the SHA-1 hash of the model. You can calculate the SHA-1 of the downloaded model and compare it with the one in the model file name by yourself (e.g. using the sha1sum command under Linux) to check the model consistency.
2. Search SSCMA Model Zoo - Issues and SSCMA - Issues to see if there are other people who have the similar problem.

If none of the above methods help, or if you have other questions about SSCMA Model Zoo, please Submit Issues to us.

Different neural networks, datasets, and models are protected by different licenses, please refer to the LICENSES for detailed permissions and restrictions.