# magicmind_edge **Repository Path**: zhaoying9150/magicmind_edge ## Basic Information - **Project Name**: magicmind_edge - **Description**: MLU边缘产品的Magicmind推理模型集合 - **Primary Language**: C++ - **License**: Apache-2.0 - **Default Branch**: master - **Homepage**: None - **GVP Project**: No ## Statistics - **Stars**: 0 - **Forks**: 29 - **Created**: 2023-01-11 - **Last Updated**: 2024-06-04 ## Categories & Tags **Categories**: Uncategorized **Tags**: None ## README # ModelZoo Edge ## 1.介绍 MagicMind是面向寒武纪 MLU 的推理加速引擎。 MagicMind能将AI框架(Tensorflow,PyTorch,ONNX,Caffe等)训练好的算法模型转换成MagicMind 统一计算图表示,并提供端到端的模型优化、代码生成以及推理业务部署能力。 本仓库展示如何将CV分类、检测、分割、NLP、语音等场景的前沿和经典模型，通过MagicMind转换和优化，进而运行在基于MagicMind的推理加速引擎的寒武纪加速板卡上的示例程序，为开发者提供丰富的AI应用移植参考。 ## 2.前提条件 - Linux 常见操作系统版本(如 Ubuntu16.04，Ubuntu18.04，CentOS7.x 等)，安装 docker(>=v18.00.0)应用程序； - 服务器装配好寒武纪 300 系列及以上的智能加速卡，并安装好驱动(>=v4.20.6)； - 若不具备以上软硬件条件，可前往[寒武纪开发者社区](https://developer.cambricon.com/)申请试用; ## 3.环境准备 若基于寒武纪云平台环境可跳过该环节。否则需运行以下步骤： 1.请前往[寒武纪开发者社区](https://developer.cambricon.com/)下载 MagicMind(version >= 0.13.0)镜像，名字如下： magicmind_version_os.tar.gz, 例如 magicmind_0.13.1-1_ubuntu18.04.tar.gz 2.加载： ```bash docker load -i magicmind_version_os.tar.gz ``` 3.运行： ```bash docker run -it --name=dockername \ --network=host --cap-add=sys_ptrace \ -v /your/host/path/MagicMind:/MagicMind \ -v /usr/bin/cnmon:/usr/bin/cnmon \ --device=/dev/cambricon_dev0:/dev/cambricon_dev0 --device=/dev/cambricon_ctl \ -w /MagicMind/ magicmind_version_image_name:tag_name /bin/bash ``` ## 4.网络支持列表和链接 ### CV： #### Detection: | MODELS | FRAMEWORK | MLU-3226 | CPP | | ------------------------------------------------------------------------ | --------- | -------- | --- | | [C3D](buildin/cv/detection/c3d_caffe) | Caffe | Yes | Yes | | [Centernet_pytorch](buildin/cv/detection/centernet_pytorch) | Pytorch | Yes | Yes | | [Mobilenet-SSD](buildin/cv/detection/mobilenet_ssd_caffe) | Caffe | Yes | Yes | | [Refinedet](buildin/cv/detection/refinedet_caffe) | Caffe | Yes | Yes | | [Retinaface](buildin/cv/detection/retinaface_pytorch) | PyTorch | Yes | Yes | | [Ultra_Fast_Lane_Detection](buildin/cv/detection/ultra_fast_lane_pytorch) | PyTorch | Yes | Yes | | [YOLOV3](buildin/cv/detection/yolov3_caffe) | Caffe | Yes | Yes | | [YOLOV3 Tiny](buildin/cv/detection/yolov3_tiny_caffe) | Caffe | Yes | Yes | | [YOLOV3_v8](buildin/cv/detection/yolov3_v8_pytorch) | PyTorch | Yes | Yes | | [YOLOV4-mish](buildin/cv/detection/yolov4_mish_caffe) | Caffe | Yes | Yes | | [YOLOV5_v6.1](buildin/cv/detection/yolov5_v6.1_pytorch) | PyTorch | Yes | Yes | | [YOLOV7](buildin/cv/detection/yolov7_pytorch) | PyTorch | Yes | Yes | #### Classification: | MODELS | FRAMEWORK | MLU-3226 | CPP | | ------------------------------------------------------------------------ | --------- | -------- | --- | | [AlexNet](buildin/cv/classification/alexnet_bn_caffe) | Caffe | Yes | Yes | | [Arcface](buildin/cv/classification/arcface_pytorch) | PyTorch | Yes | Yes | | [DenseNet121](buildin/cv/classification/densenet121_caffe) | Caffe | Yes | Yes | | [DenseNet201](buildin/cv/classification/densenet201_caffe) | Caffe | Yes | Yes | | [Googlenet_bn](buildin/cv/classification/googlenet_bn_caffe) | Caffe | Yes | Yes | | [Inceptionv2](buildin/cv/classification/inceptionv2_caffe) | Caffe | Yes | Yes | | [Inceptionv3](buildin/cv/classification/inceptionv3_caffe) | Caffe | Yes | Yes | | [Inceptionv4](buildin/cv/classification/inceptionv4_caffe) | Caffe | Yes | Yes | | [Mobilenet-SSD](buildin/cv/detection/mobilenet_ssd_caffe) | Caffe | Yes | Yes | | [Mobilenetv2](buildin/cv/classification/mobilenetv2_caffe) | Caffe | Yes | Yes | | [Mobilenetv3](buildin/cv/classification/mobilenetv3_pytorch) | Pytorch | Yes | Yes | | [Resnet50](buildin/cv/classification/resnet50_caffe) | Caffe | Yes | Yes | | [Resnext50](buildin/cv/classification/resnext50_caffe) | Caffe | Yes | Yes | | [Senet50](buildin/cv/classification/senet50_caffe) | Caffe | Yes | Yes | | [Squeezenet1.0](buildin/cv/classification/squeezenet_v1_0_caffe) | Caffe | Yes | Yes | | [Squeezenet1.1](buildin/cv/classification/squeezenet_v1_1_caffe) | Caffe | Yes | Yes | | [VGG16](buildin/cv/classification/vgg16_caffe) | Caffe | Yes | Yes | #### Segmentation: | MODELS | FRAMEWORK | MLU-3226 | CPP | | ------------------------------------------------------------------------ | --------- | -------- | --- | | [DeepLabv3](buildin/cv/segmentation/deeplabv3_tf) | Tensorflow| Yes | Yes | | [nnUNet](buildin/cv/segmentation/nnUNet_pytorch) | Pytorch | Yes | Yes | | [SegNet](buildin/cv/segmentation/segnet_caffe) | Caffe | Yes | Yes | #### OCR: | MODELS | FRAMEWORK | MLU-3226 | CPP | | ------------------------------------------------------------------------ | --------- | -------- | --- | | [CRNN](buildin/cv/ocr/crnn_pytorch) | Pytorch | Yes | Yes | | [DBnet](buildin/cv/ocr/dbnet_pytorch) | Pytorch | Yes | Yes | ### Others: | MODELS | FRAMEWORK | MLU-3226 | CPP | | ------------------------------------------------------------------------ | --------- | -------- | --- | | [Openpose](buildin/cv/other/opense_caffe) | Caffe | Yes | Yes | ## 5.issues/wiki/forum 跳转链接 ## 6.contrib 指引和链接 ## 7.LICENSE ModelZoo Edge 的 License 具体内容请参见[LICENSE](LICENSE)文件。 ## 8.免责声明 ModelZoo 仅提供公共数据集以及预训练模型的下载链接，公共数据集及预训练模型并不属于 ModelZoo, ModelZoo 也不对其质量或维护承担责任。请您在使用公共数据集和预训练模型的过程中，确保符合其对应的使用许可。 如果您不希望您的数据集或模型公布在 ModelZoo 上，或者您希望更新 ModelZoo 中属于您的数据集或模型，请您通过 Gitee 中提交 issue ，您也可以联系 ecosystem@cambricon.com 告知我们。 # Release notes ## v1.2 ### 新增内容 1. 新增 CRNN 网络MagicMind edge支持 2. 新增 DBnet 网络MagicMind edge支持 3. 新增 Retinaface 网络MagicMind edge支持 4. 新增 Refinedet 网络MagicMind edge支持 5. 新增 Openpose 网络MagicMind edge支持 6. 新增 Ultra-Fast-Lane-Detection 网络MagicMind edge支持 7. 新增 Googlenet 网络MagicMind edge支持 8. 新增 YOLOV7 网络MagicMind edge支持 ### Bug fix 1. 修复YOLOV5 由前后处理导致的精度问题 2. 修复所有模型 README.MD 中描述问题以及一键运行脚本相关bug ## v1.1 ### 新增内容 1. CV:ALexNet_with_bn_caffe, Arcface, C3D, Centernet, DeepLabv3, DensetNet121/201, Inceptionv2/3/4, Mobilenet-SSD, Mobilenetv2/v3, nnUNet, Resnet50, Resnext50, SegNet, Senet50, Squeezenet1.0/1.1, VGG16, YOLOV5_v6.1, YOLOV3_v8, YOLOV3, YOLOV4-mish 共计25个模型.