The Vitis™ AI Model Zoo includes optimized deep learning models to speed up the deployment of deep learning inference applications on Xilinx® platforms. These models cover different application fields, including but not limited to ADAS/AD, video surveillance, robotics, data center, etc. You can get started with these free pre-trained models to enjoy the benefits of deep learning acceleration.
In Vitis 1.3 AI Model Zoo, a variety of Neural Network models with three popular frameworks, Caffe, TensorFlow and PyTorch, are provided. For every model, a .yaml file that provides a description of model name, framework, task type, network backbone, train & validation dataset, float OPS, prune or not, download link, license, and md5 checksum is released. You can browse a model list in Vitis 1.3 AI Model Zoo and select a Neural Network model that you are interested in and get its basic information from a specified .yaml file. With the download link in the .yaml file, you can download the model freely.
For example, if you need a ResNet-50 model used for general image classification on TensorFlow, then find a model named tf_resnetv1_50_imagenet_224_224_6.97G_1.3. According to standard naming rules, models are named using this format: F_M_(D)_H_W_(P)_C_V.
- F specifies training framework: cf is caffe, tf is TensorFlow, dk is Darknet, pt is PyTorch.
- M specifies the model feature.
- D specifies the dataset. It is optional depending on whether the dataset is public or private. Mixed means a mixture of multiple public datasets.
- H specifies the height of input data.
- W specifies the width of input data.
- P specifies the pruning ratio, it means how much computation is reduced. It is optional depending on whether the model is pruned or not.
- C specifies the computation of the model: how many Gops per image.
- V specifies the version of Vitis AI.
As such, tf_resnetv1_50_imagenet_224_224_6.97G_1.3 is a ResNet v1-50 model trained with TensorFlow using the Imagenet dataset, input data size is 224*224, not pruned, the computation per image is 6.97 Gops and Vitis AI version is 1.3.
Then you can choose this model and download it manually using the link provided in tf_resnetv1_50_imagenet_224_224_6.97G_1.3.yaml or through tools that can read .yaml information.
For more information about models list, see https://github.com/Xilinx/Vitis-AI/tree/master/models/AI-Model-Zoo/model-list.