--- library_name: pytorch license: other tags: - backbone - android pipeline_tag: image-classification --- ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/regnet/web-assets/model_demo.png) # RegNet: Optimized for Mobile Deployment ## Imagenet classifier and general purpose backbone RegNet is a machine learning model that can classify images from the Imagenet dataset. It can also be used as a backbone in building more complex models for specific use cases. This model is an implementation of RegNet found [here](https://github.com/pytorch/vision/blob/main/torchvision/models/regnet.py). This repository provides scripts to run RegNet on Qualcomm® devices. More details on model performance across various devices, can be found [here](https://aihub.qualcomm.com/models/regnet). ### Model Details - **Model Type:** Model_use_case.image_classification - **Model Stats:** - Model checkpoint: Imagenet - Input resolution: 224x224 - Number of parameters: 15.3M - Model size (float): 58.3 MB - Model size (w8a8): 15.4 MB | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model |---|---|---|---|---|---|---|---|---| | RegNet | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 9.973 ms | 0 - 67 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.tflite) | | RegNet | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 9.919 ms | 1 - 47 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.dlc) | | RegNet | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 2.745 ms | 0 - 72 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.tflite) | | RegNet | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 3.315 ms | 1 - 51 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.dlc) | | RegNet | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 2.005 ms | 0 - 84 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.tflite) | | RegNet | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 1.992 ms | 1 - 13 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.dlc) | | RegNet | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 2.031 ms | 1 - 14 MB | NPU | [RegNet.onnx.zip](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.onnx.zip) | | RegNet | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 3.244 ms | 0 - 67 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.tflite) | | RegNet | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 3.255 ms | 0 - 47 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.dlc) | | RegNet | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 9.973 ms | 0 - 67 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.tflite) | | RegNet | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 9.919 ms | 1 - 47 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.dlc) | | RegNet | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 2.006 ms | 0 - 215 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.tflite) | | RegNet | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 2.006 ms | 1 - 17 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.dlc) | | RegNet | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 3.397 ms | 0 - 64 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.tflite) | | RegNet | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 3.443 ms | 1 - 47 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.dlc) | | RegNet | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 2.008 ms | 0 - 203 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.tflite) | | RegNet | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 2.014 ms | 1 - 15 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.dlc) | | RegNet | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 3.244 ms | 0 - 67 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.tflite) | | RegNet | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 3.255 ms | 0 - 47 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.dlc) | | RegNet | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 1.392 ms | 0 - 82 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.tflite) | | RegNet | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 1.379 ms | 1 - 59 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.dlc) | | RegNet | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 1.469 ms | 0 - 56 MB | NPU | [RegNet.onnx.zip](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.onnx.zip) | | RegNet | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 1.141 ms | 0 - 71 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.tflite) | | RegNet | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 1.085 ms | 1 - 54 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.dlc) | | RegNet | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 1.164 ms | 0 - 51 MB | NPU | [RegNet.onnx.zip](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.onnx.zip) | | RegNet | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | TFLITE | 0.968 ms | 0 - 69 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.tflite) | | RegNet | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | QNN_DLC | 0.9 ms | 0 - 53 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.dlc) | | RegNet | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | ONNX | 1.017 ms | 0 - 51 MB | NPU | [RegNet.onnx.zip](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.onnx.zip) | | RegNet | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 2.214 ms | 110 - 110 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.dlc) | | RegNet | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 1.943 ms | 39 - 39 MB | NPU | [RegNet.onnx.zip](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet.onnx.zip) | | RegNet | w8a8 | Dragonwing RB3 Gen 2 Vision Kit | Qualcomm® QCS6490 | TFLITE | 2.297 ms | 0 - 21 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.tflite) | | RegNet | w8a8 | Dragonwing RB3 Gen 2 Vision Kit | Qualcomm® QCS6490 | QNN_DLC | 2.502 ms | 0 - 134 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.dlc) | | RegNet | w8a8 | Dragonwing RB3 Gen 2 Vision Kit | Qualcomm® QCS6490 | ONNX | 27.765 ms | 8 - 18 MB | CPU | [RegNet.onnx.zip](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.onnx.zip) | | RegNet | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 1.992 ms | 0 - 47 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.tflite) | | RegNet | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 2.277 ms | 0 - 48 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.dlc) | | RegNet | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 0.921 ms | 0 - 71 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.tflite) | | RegNet | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 1.284 ms | 0 - 63 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.dlc) | | RegNet | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 0.76 ms | 0 - 50 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.tflite) | | RegNet | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 0.891 ms | 0 - 14 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.dlc) | | RegNet | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 1.217 ms | 0 - 14 MB | NPU | [RegNet.onnx.zip](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.onnx.zip) | | RegNet | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 1.101 ms | 0 - 47 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.tflite) | | RegNet | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 1.27 ms | 0 - 48 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.dlc) | | RegNet | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | TFLITE | 51.11 ms | 3 - 87 MB | GPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.tflite) | | RegNet | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | ONNX | 17.591 ms | 5 - 15 MB | CPU | [RegNet.onnx.zip](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.onnx.zip) | | RegNet | w8a8 | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 1.992 ms | 0 - 47 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.tflite) | | RegNet | w8a8 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 2.277 ms | 0 - 48 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.dlc) | | RegNet | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 0.754 ms | 0 - 49 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.tflite) | | RegNet | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 0.911 ms | 0 - 15 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.dlc) | | RegNet | w8a8 | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 1.368 ms | 0 - 50 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.tflite) | | RegNet | w8a8 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 1.589 ms | 0 - 51 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.dlc) | | RegNet | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 0.771 ms | 0 - 51 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.tflite) | | RegNet | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 0.91 ms | 0 - 27 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.dlc) | | RegNet | w8a8 | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 1.101 ms | 0 - 47 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.tflite) | | RegNet | w8a8 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 1.27 ms | 0 - 48 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.dlc) | | RegNet | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 0.544 ms | 0 - 73 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.tflite) | | RegNet | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 0.659 ms | 0 - 70 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.dlc) | | RegNet | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 0.796 ms | 0 - 77 MB | NPU | [RegNet.onnx.zip](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.onnx.zip) | | RegNet | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 0.432 ms | 0 - 55 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.tflite) | | RegNet | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 0.482 ms | 0 - 55 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.dlc) | | RegNet | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 0.692 ms | 0 - 54 MB | NPU | [RegNet.onnx.zip](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.onnx.zip) | | RegNet | w8a8 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | TFLITE | 1.032 ms | 0 - 52 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.tflite) | | RegNet | w8a8 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | QNN_DLC | 1.173 ms | 0 - 54 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.dlc) | | RegNet | w8a8 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | ONNX | 17.035 ms | 8 - 26 MB | CPU | [RegNet.onnx.zip](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.onnx.zip) | | RegNet | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | TFLITE | 0.42 ms | 1 - 48 MB | NPU | [RegNet.tflite](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.tflite) | | RegNet | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | QNN_DLC | 0.435 ms | 0 - 54 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.dlc) | | RegNet | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | ONNX | 0.671 ms | 0 - 55 MB | NPU | [RegNet.onnx.zip](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.onnx.zip) | | RegNet | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 1.075 ms | 66 - 66 MB | NPU | [RegNet.dlc](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.dlc) | | RegNet | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 1.123 ms | 20 - 20 MB | NPU | [RegNet.onnx.zip](https://huggingface.co/qualcomm/RegNet/blob/main/RegNet_w8a8.onnx.zip) | ## Installation Install the package via pip: ```bash pip install qai-hub-models ``` ## Configure Qualcomm® AI Hub Workbench to run this model on a cloud-hosted device Sign-in to [Qualcomm® AI Hub Workbench](https://workbench.aihub.qualcomm.com/) with your Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`. With this API token, you can configure your client to run models on the cloud hosted devices. ```bash qai-hub configure --api_token API_TOKEN ``` Navigate to [docs](https://workbench.aihub.qualcomm.com/docs/) for more information. ## Demo off target The package contains a simple end-to-end demo that downloads pre-trained weights and runs this model on a sample input. ```bash python -m qai_hub_models.models.regnet.demo ``` The above demo runs a reference implementation of pre-processing, model inference, and post processing. **NOTE**: If you want running in a Jupyter Notebook or Google Colab like environment, please add the following to your cell (instead of the above). ``` %run -m qai_hub_models.models.regnet.demo ``` ### Run model on a cloud-hosted device In addition to the demo, you can also run the model on a cloud-hosted Qualcomm® device. This script does the following: * Performance check on-device on a cloud-hosted device * Downloads compiled assets that can be deployed on-device for Android. * Accuracy check between PyTorch and on-device outputs. ```bash python -m qai_hub_models.models.regnet.export ``` ## How does this work? This [export script](https://aihub.qualcomm.com/models/regnet/qai_hub_models/models/RegNet/export.py) leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model on-device. Lets go through each step below in detail: Step 1: **Compile model for on-device deployment** To compile a PyTorch model for on-device deployment, we first trace the model in memory using the `jit.trace` and then call the `submit_compile_job` API. ```python import torch import qai_hub as hub from qai_hub_models.models.regnet import Model # Load the model torch_model = Model.from_pretrained() # Device device = hub.Device("Samsung Galaxy S25") # Trace model input_shape = torch_model.get_input_spec() sample_inputs = torch_model.sample_inputs() pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()]) # Compile model on a specific device compile_job = hub.submit_compile_job( model=pt_model, device=device, input_specs=torch_model.get_input_spec(), ) # Get target model to run on-device target_model = compile_job.get_target_model() ``` Step 2: **Performance profiling on cloud-hosted device** After compiling models from step 1. Models can be profiled model on-device using the `target_model`. Note that this scripts runs the model on a device automatically provisioned in the cloud. Once the job is submitted, you can navigate to a provided job URL to view a variety of on-device performance metrics. ```python profile_job = hub.submit_profile_job( model=target_model, device=device, ) ``` Step 3: **Verify on-device accuracy** To verify the accuracy of the model on-device, you can run on-device inference on sample input data on the same cloud hosted device. ```python input_data = torch_model.sample_inputs() inference_job = hub.submit_inference_job( model=target_model, device=device, inputs=input_data, ) on_device_output = inference_job.download_output_data() ``` With the output of the model, you can compute like PSNR, relative errors or spot check the output with expected output. **Note**: This on-device profiling and inference requires access to Qualcomm® AI Hub Workbench. [Sign up for access](https://myaccount.qualcomm.com/signup). ## Run demo on a cloud-hosted device You can also run the demo on-device. ```bash python -m qai_hub_models.models.regnet.demo --eval-mode on-device ``` **NOTE**: If you want running in a Jupyter Notebook or Google Colab like environment, please add the following to your cell (instead of the above). ``` %run -m qai_hub_models.models.regnet.demo -- --eval-mode on-device ``` ## Deploying compiled model to Android The models can be deployed using multiple runtimes: - TensorFlow Lite (`.tflite` export): [This tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a guide to deploy the .tflite model in an Android application. - QNN (`.so` export ): This [sample app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html) provides instructions on how to use the `.so` shared library in an Android application. ## View on Qualcomm® AI Hub Get more details on RegNet's performance across various devices [here](https://aihub.qualcomm.com/models/regnet). Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/) ## License * The license for the original implementation of RegNet can be found [here](https://github.com/pytorch/vision/blob/main/LICENSE). * The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf) ## References * [Designing Network Design Spaces](https://arxiv.org/abs/2003.13678) * [Source Model Implementation](https://github.com/pytorch/vision/blob/main/torchvision/models/regnet.py) ## Community * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI. * For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).