# ModelZoo **Repository Path**: HiSpark/modelzoo ## Basic Information - **Project Name**: ModelZoo - **Description**: ModelZoo,HiSpark下的开源AI模型平台,涵盖计算机视觉、自然语言处理、语音、推荐、多模态、大语言模型等方向的AI模型及其基于海思实操案例。 - **Primary Language**: C++ - **License**: Apache-2.0 - **Default Branch**: master - **Homepage**: None - **GVP Project**: No ## Statistics - **Stars**: 4 - **Forks**: 6 - **Created**: 2025-07-21 - **Last Updated**: 2025-12-05 ## Categories & Tags **Categories**: Uncategorized **Tags**: None ## README # ModelZoo # 简介 ModelZoo,HiSpark下的开源AI模型平台,涵盖计算机视觉、自然语言处理、语音、推荐、多模态等方向的AI模型。我们仅对模型做适配和格式转化,未重新训练模型和对模型进行功能性修改。我们提供模型基于海思实操案例demo和性能表现,供开发者学习。平台的每个模型都有详细的使用指导,为方便更多开发者使用ModelZoo,我们将持续增加典型网络和相关预训练模型。如果您有任何需求,请在 **Gitee** 提交 [**issue**](https://gitee.com/HiSpark/modelzoo/issues) ,我们会及时处理。 # 目录 | 目录 | 说明 | | -------- | ------- | | docs | 文档说明 | | datasets | 数据集 | | samples | 模型 | | utils | 工具 | # 模型列表 ## 运行用户建议 **说明:** **因使用版本差异,模型性能可能存在波动,性能仅供参考**
模型 数据集 SVP NNN性能fps NNN性能fps 输入
SqueezeNet1_1 ImageNet 2052.42 801.62 224 x 224
ShuffleNetV2 ImageNet 403.39 255 224 x 224
ResNet50 ImageNet 336.66 133.10 224 x 224
MobileNetV2 ImageNet 1317.222 311.80 224 x 224
EfficientNetV2 ImageNet 37.051 27.338 288 x 288
DenseNet121 ImageNet 170.63 108.01 224 x 224
Swin-Transformer ImageNet 38.24 9.81 224 x 224
Vit-B-16 ImageNet 28.56 6.85 224 x 224
ResNet101 ImageNet 210.675 83.102 224 x 224
ResNet18 ImageNet 753.02 500 x 375
Chinese-CLIP CIFAR100 11.142 224 x 224
Depth-Anything-v2 DA-2K 3.756 518 x 518
PFLD WFLW 1631.88 112 x 112
SuperPoint HPatches 320.60 240 x 320
UNet carvana 9.6372 2.14 572 x 572
YOLOv5s coco2017 85.41 35.54 640 x 640
YOLOv8l coco2017 10.393 640 x 640
YOLOv8s coco2017 42.914 640 x 640
YOLO11s coco2017 42.753 640 x 640
YOLO11s-seg coco2017 32.732 640 x 640
InceptionV3 ImageNet 192.42 95.48 299 x 299
YOLOv3 coco2017 17.327 640 x 640
YOLOv7 coco2017 13.45 7.65 640 x 640
YOLOv4 coco2017 3.84 608 x 608
YOLO11s-pose coco2017 38.6 640 x 640
YOLOv6s coco2017 39.51 640 x 640
SE-ResNet50 ImageNet 13.56 63.83 224 x 224
VGG16 ImageNet 72.27 32.53 224 x 224
CodeFormer CelebA-512 1.38 512 x 512
# 模型适配AI引擎计划 **注:** **适配计划如下所示,具体时间节点,以实际发布为准** ## Hi3403
| *AI引擎* | **2025Q3** | **2025Q4** | **2026Q1** | **2026Q2** | **2026Q3** | | :------: | :----------------------------------------------------------: | :----------------------------------------------------------: | :--------------------: | :-----------------: | :--------------------: | | SVP NNN | [**SqueezeNet1_1**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/SqueezeNet1_1) | [**VGG16**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/vgg16) | **YOLOV8s-OBB** | | | | | [**ShuffleNetV2**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/ShuffleNetV2) | [**InceptionV3**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/InceptionV3) | **YOLOV9s** | | | | | [**ResNet50**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/ResNet50) | [**SE-ResNet50**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/SEResNet50) | **Yolov8s-world** | | | | | [**MobileNetV2**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/MobileNetV2) | [**YOLOV3**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/samples_GPL/built-in/yolov3) | **centernet** | | | | | [**EfficientNetV2**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/EfficientNetV2) | [**YOLOV4**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/detection/yolov4) | **deepsort** | | | | | [**DenseNet121**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/DenseNet121) | [**YOLOV6s**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/samples_GPL/built-in/yolov6s) | **x-Stereo** | | | | | [**Swin-Transformer**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/Swin-Transformer) | [**YOLOv7**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/samples_GPL/built-in/yolov7) | **Yolov8s-seg** | | | | | [**Vit-B-16**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/Vit-B-16) | **YOLOv10s** | **tinysan(MobileSam)** | | | | | [**ResNet101**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/ResNet101) | [**YOLO11s-pose**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/samples_GPL/built-in/yolo11s-pose) | **Siamese network** | | | | | [**ResNet18**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/ResNet18) | **HRNet** | **Facenet** | | | | | [**Chinese-CLIP**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/Chinese-CLIP) | **CrowdCount** | **PaddleOCR-rec** | | | | | [**Depth-Anything-v2**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/depth/Depth-Anything-v2) | [**CodeFormer**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/super_resolution/CodeFormer) | **PaddleOCR-det** | | | | | [**PFLD**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/detection/PFLD) | | **CRNN** | | | | | [**SuperPoint**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/point/SuperPoint) | | **VDSR** | | | | | [**UNet**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/samples_GPL/built-in/UNet) | | **FastSpeech2** | | | | | [**YOLOv5s**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/samples_GPL/built-in/yolov5) | | **minicpm(VLM 0.5B)** | | | | | [**YOLOv8l**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/samples_GPL/built-in/yolov8l) | | | | | | | [**YOLOv8s**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/samples_GPL/built-in/yolov8s) | | | | | | | [**YOLO11s**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/samples_GPL/built-in/yolo11s) | | | | | | | [**YOLO11s-seg**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/samples_GPL/built-in/yolo11s-seg) | | | | | | NNN | [**SqueezeNet1_1**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/SqueezeNet1_1) | [**VGG16**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/vgg16) | **CN-CLIP** | **YOLOV8s-OBB** | **YOLOV10s** | | | [**ShuffleNetV2**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/ShuffleNetV2) | [**InceptionV3**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/InceptionV3) | **YOLOV3** | **centernet** | **Yolov8s-world** | | | [**ResNet50**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/ResNet50) | [**SE-ResNet50**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/SEResNet50) | **YOLOV4** | **deepsort** | **tinysan(MobileSam)** | | | [**MobileNetV2**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/MobileNetV2) | [**EfficientNetV2**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/EfficientNetV2) | **YOLOV8s** | **DepthAnythingV2** | | | | [**DenseNet121**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/DenseNet121) | [**Swin-Transformer**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/Swin-Transformer) | **YOLOV9s** | **Siamese network** | | | | [**ResNet101**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/ResNet101) | [**Vit-B-16**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/built-in/classification/Vit-B-16) | **YOLOV11s** | **Facenet** | | | | [**YOLOv5s**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/samples_GPL/built-in/yolov5) | **YOLOV6s** | **YOLO11s-seg** | **PaddleOCR-rec** | | | | | [**YOLOv7**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/samples_GPL/built-in/yolov7) | **YOLO11s-pose** | **PaddleOCR-det** | | | | | [**UNet**](https://gitee.com/HiSpark/modelzoo/tree/master/samples/samples_GPL/built-in/UNet) | **Yolov8s-seg** | **CRNN** | | | | | **HRNet** | **PFLD** | **VDSR** | | | | | **CrowdCount** | **CodeFormer** | **FastSpeech2** | | | | | | **SuperPoint** | | |
## Hi3591P
| **2026Q1** | **pi0** | **graspnet** | |:----------:|:-------:|:------------:|
# 如何贡献 本仓子模块参考目录,可以直接克隆子仓,也可以克隆主仓,在开始贡献之前,请先阅读[NOTICE](https://gitee.com/HiSpark/docs/blob/master/contribute/%E7%A4%BE%E5%8C%BA%E5%8F%82%E4%B8%8E%E8%B4%A1%E7%8C%AE%E6%8C%87%E5%8D%97.md),谢谢! # 免责声明 ## 致ModelZoo使用者 * HiSpark ModelZoo提供的模型仅供您用于非商业目的。 * HiSpark ModelZoo仅提供公共数据集下载、模型下载和预处理脚本。这些数据集和模型不属于ModelZoo,ModelZoo也不对其质量或维护负责。请确保您具有这些数据集和模型的使用许可,如您因使用数据集和模型产生侵权纠纷,海思不承担任何责任。 * 如您在使用ModelZoo模型过程中,发现任何问题(包括但不限于功能问题、合规问题),请在Gitee提交issue,我们将及时审视并解决。 ## 致数据集、模型所有者 如果您不希望您的数据集、模型公布在ModelZoo上或希望更新ModelZoo中属于您的数据集、模型,请在Gitee提交issue,我们将根据您的issue删除或更新您的数据集、模型。衷心感谢您对ModelZoo的理解和贡献。 ## License声明 HiSpark ModelZoo提供的模型,如模型目录下存在License的,以该License为准。如模型目录下不存在License的,以Apache 2.0许可证许可,对应许可证文本可查阅HiSpark ModelZoo根目录。