--- title: MaskRCNN canonical_url: "https://www.modelscope.cn/models/Qualcomm-AI-Hub/MaskRCNN" md_url: "https://www.modelscope.cn/models/Qualcomm-AI-Hub/MaskRCNN.md" repository: Qualcomm-AI-Hub/MaskRCNN last_updated: 2026-10-16 license: bsd-3-clause pipeline_tag: image-segmentation tasks: - image-segmentation downloads: 0 stars: 0 tags: - android --- # MaskRCNN > MaskRCNN - Qualcomm-AI-Hub 在 ModelScope 开源的模型。MaskRCNN: Optimized for Qualcomm Devices Qualcomm-AI-Hub/MaskRCNN 是 ModelScope 魔搭社区上的image-segmentation模型，采用 bsd-3-clause 许可。 - **Repository**: Qualcomm-AI-Hub/MaskRCNN - **License**: bsd-3-clause - **Tasks**: image-segmentation - **Tags**: android - **Downloads**: 0 - **Stars**: 0 - **Last updated**: 2026-10-16 Source: https://www.modelscope.cn/models/Qualcomm-AI-Hub/MaskRCNN --- ![](https://china-qaihub-public-assets.s3.cn-northwest-1.amazonaws.com.cn/qai-hub-models/models/maskrcnn/web-assets/model_demo.png) # MaskRCNN: Optimized for Qualcomm Devices Mask R-CNN is a machine learning model that extends Faster R-CNN to perform instance segmentation by detecting objects in an image while simultaneously generating a high-quality segmentation mask for each instance. It adds a branch for predicting segmentation masks in parallel with the existing branch for bounding box recognition. This is based on the implementation of MaskRCNN found [here](https://github.com/pytorch/vision). This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models-cn/blob/main/src/qai_hub_models/models/maskrcnn) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary). Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.cn) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device. ## Getting Started There are two ways to deploy this model on your device: ### Option 1: Download Pre-Exported Models Below are pre-exported model assets ready for deployment. | Runtime | Precision | Chipset | SDK Versions | Download | |---|---|---|---|---| | ONNX | float | Universal | QAIRT 2.45, ONNX Runtime 1.27.1 | [Download](https://china-qaihub-public-assets.s3.cn-northwest-1.amazonaws.com.cn/qai-hub-models/models/maskrcnn/releases/v0.62.1/maskrcnn-onnx-float.zip) | QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://china-qaihub-public-assets.s3.cn-northwest-1.amazonaws.com.cn/qai-hub-models/models/maskrcnn/releases/v0.62.1/maskrcnn-qnn_dlc-float.zip) For more device-specific assets and performance metrics, visit **[MaskRCNN on Qualcomm® AI Hub](https://aihub.qualcomm.cn/models/maskrcnn)**. ### Option 2: Export with Custom Configurations Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models-cn/blob/main/src/qai_hub_models/models/maskrcnn) Python library to compile and export the model with your own: - Custom weights (e.g., fine-tuned checkpoints) - Custom input shapes - Target device and runtime configurations This option is ideal if you need to customize the model beyond the default configuration provided here. See our repository for [MaskRCNN on GitHub](https://github.com/qualcomm/ai-hub-models-cn/blob/main/src/qai_hub_models/models/maskrcnn) for usage instructions. ## Model Details **Model Type:** Model_use_case.semantic_segmentation **Model Stats:** - Input resolution: 800x800 - Model checkpoint: Mask R-CNN ResNet-50 FPN V2 - Model size (float): 177 MB - Number of output classes: 91 - Number of parameters: 46.4M ## Performance Summary | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit |---|---|---|---|---|---|--- | proposal_generator | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 70.231 ms | 0 - 2812 MB | NPU | proposal_generator | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 142.534 ms | 130 - 2863 MB | NPU | proposal_generator | ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 124.965 ms | 24 - 35 MB | NPU | proposal_generator | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 92.531 ms | 123 - 199 MB | NPU | proposal_generator | ONNX | float | Qualcomm® QCS8450 | 142.534 ms | 130 - 2863 MB | NPU | proposal_generator | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 119.396 ms | 7 - 17 MB | NPU | proposal_generator | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 52.601 ms | 127 - 1623 MB | NPU | proposal_generator | ONNX | float | Snapdragon® 8 Elite Mobile | 52.601 ms | 127 - 1623 MB | NPU | proposal_generator | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 41.473 ms | 126 - 1479 MB | NPU | proposal_generator | QNN_DLC | float | Snapdragon® X2 Elite | 42.762 ms | 7 - 7 MB | NPU | proposal_generator | QNN_DLC | float | Snapdragon® X Elite | 92.038 ms | 7 - 7 MB | NPU | proposal_generator | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 69.972 ms | 7 - 2330 MB | NPU | proposal_generator | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 147.799 ms | 8 - 2739 MB | NPU | proposal_generator | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 127.626 ms | 7 - 72 MB | NPU | proposal_generator | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 94.57 ms | 7 - 10 MB | NPU | proposal_generator | QNN_DLC | float | Qualcomm® SA8775P | 122.238 ms | 1 - 1758 MB | NPU | proposal_generator | QNN_DLC | float | Qualcomm® SA8650P | 122.238 ms | 1 - 1758 MB | NPU | proposal_generator | QNN_DLC | float | Qualcomm® SA8255P | 122.238 ms | 1 - 1758 MB | NPU | proposal_generator | QNN_DLC | float | Qualcomm® QCS8450 | 147.799 ms | 8 - 2739 MB | NPU | proposal_generator | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 122.505 ms | 7 - 71 MB | NPU | proposal_generator | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 92.038 ms | 7 - 7 MB | NPU | proposal_generator | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 53.247 ms | 0 - 1510 MB | NPU | proposal_generator | QNN_DLC | float | Qualcomm® SA7255P | 353.632 ms | 2 - 1759 MB | NPU | proposal_generator | QNN_DLC | float | Qualcomm® SA8295P | 125.972 ms | 0 - 1430 MB | NPU | proposal_generator | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 53.247 ms | 0 - 1510 MB | NPU | proposal_generator | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 42.128 ms | 7 - 1789 MB | NPU | roi_head | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 72.432 ms | 244 - 718 MB | NPU | roi_head | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 138.03 ms | 234 - 680 MB | NPU | roi_head | ONNX | float | Qualcomm® Dragonwing™ IQ-8275 | 154.653 ms | 108 - 208 MB | NPU | roi_head | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 87.331 ms | 271 - 273 MB | NPU | roi_head | ONNX | float | Qualcomm® QCS8450 | 138.03 ms | 234 - 680 MB | NPU | roi_head | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 109.801 ms | 112 - 211 MB | NPU | roi_head | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 62.278 ms | 226 - 623 MB | NPU | roi_head | ONNX | float | Snapdragon® 8 Elite Mobile | 62.278 ms | 226 - 623 MB | NPU | roi_head | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 50.521 ms | 190 - 588 MB | NPU | roi_head | QNN_DLC | float | Snapdragon® X2 Elite | 89.841 ms | 52 - 52 MB | NPU | roi_head | QNN_DLC | float | Snapdragon® X Elite | 231.039 ms | 52 - 52 MB | NPU | roi_head | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 159.1 ms | 10 - 795 MB | NPU | roi_head | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 295.243 ms | 39 - 787 MB | NPU | roi_head | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-8275 | 243.435 ms | 52 - 107 MB | NPU | roi_head | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 224.388 ms | 51 - 52 MB | NPU | roi_head | QNN_DLC | float | Qualcomm® SA8775P | 261.792 ms | 49 - 777 MB | NPU | roi_head | QNN_DLC | float | Qualcomm® SA8650P | 261.792 ms | 49 - 777 MB | NPU | roi_head | QNN_DLC | float | Qualcomm® SA8255P | 261.792 ms | 49 - 777 MB | NPU | roi_head | QNN_DLC | float | Qualcomm® QCS8450 | 295.243 ms | 39 - 787 MB | NPU | roi_head | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 256.185 ms | 52 - 106 MB | NPU | roi_head | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 231.039 ms | 52 - 52 MB | NPU | roi_head | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 116.843 ms | 1 - 490 MB | NPU | roi_head | QNN_DLC | float | Qualcomm® SA7255P | 560.942 ms | 31 - 534 MB | NPU | roi_head | QNN_DLC | float | Qualcomm® SA8295P | 287.976 ms | 49 - 769 MB | NPU | roi_head | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 116.843 ms | 1 - 490 MB | NPU | roi_head | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 82.518 ms | 13 - 513 MB | NPU ## License * The license for the original implementation of MaskRCNN can be found [here](https://github.com/pytorch/vision/blob/main/LICENSE). ## References * [Mask R-CNN](https://arxiv.org/abs/1703.06870) * [Source Model Implementation](https://github.com/pytorch/vision) ## Community * For questions or feedback, chat with our engineers in [Qualcomm® AI Hub Workbench](https://workbench.aihub.qualcomm.cn).