--- library_name: pytorch license: other tags: - backbone - real_time - android pipeline_tag: image-segmentation --- ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/unet_segmentation/web-assets/model_demo.png) # Unet-Segmentation: Optimized for Mobile Deployment ## Real-time segmentation optimized for mobile and edge UNet is a machine learning model that produces a segmentation mask for an image. The most basic use case will label each pixel in the image as being in the foreground or the background. More advanced usage will assign a class label to each pixel. This version of the model was trained on the data from Kaggle's Carvana Image Masking Challenge (see https://www.kaggle.com/c/carvana-image-masking-challenge) and is used for vehicle segmentation. This model is an implementation of Unet-Segmentation found [here](https://github.com/milesial/Pytorch-UNet). This repository provides scripts to run Unet-Segmentation on Qualcomm® devices. More details on model performance across various devices, can be found [here](https://aihub.qualcomm.com/models/unet_segmentation). ### Model Details - **Model Type:** Model_use_case.semantic_segmentation - **Model Stats:** - Model checkpoint: unet_carvana_scale1.0_epoch2 - Input resolution: 224x224 - Number of output classes: 2 (foreground / background) - Number of parameters: 31.0M - Model size (float): 118 MB - Model size (w8a8): 29.8 MB | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model |---|---|---|---|---|---|---|---|---| | Unet-Segmentation | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 952.953 ms | 0 - 98 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.tflite) | | Unet-Segmentation | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 946.447 ms | 0 - 110 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.dlc) | | Unet-Segmentation | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 266.695 ms | 6 - 162 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.tflite) | | Unet-Segmentation | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 279.553 ms | 9 - 130 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.dlc) | | Unet-Segmentation | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 156.062 ms | 6 - 465 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.tflite) | | Unet-Segmentation | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 145.554 ms | 9 - 54 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.dlc) | | Unet-Segmentation | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 253.329 ms | 6 - 104 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.tflite) | | Unet-Segmentation | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 249.775 ms | 0 - 110 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.dlc) | | Unet-Segmentation | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 952.953 ms | 0 - 98 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.tflite) | | Unet-Segmentation | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 946.447 ms | 0 - 110 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.dlc) | | Unet-Segmentation | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 158.946 ms | 3 - 467 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.tflite) | | Unet-Segmentation | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 149.352 ms | 9 - 54 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.dlc) | | Unet-Segmentation | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 280.575 ms | 6 - 107 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.tflite) | | Unet-Segmentation | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 276.093 ms | 2 - 113 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.dlc) | | Unet-Segmentation | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 152.13 ms | 6 - 472 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.tflite) | | Unet-Segmentation | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 152.947 ms | 10 - 54 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.dlc) | | Unet-Segmentation | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 253.329 ms | 6 - 104 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.tflite) | | Unet-Segmentation | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 249.775 ms | 0 - 110 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.dlc) | | Unet-Segmentation | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 151.859 ms | 6 - 466 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.tflite) | | Unet-Segmentation | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 151.338 ms | 9 - 55 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.dlc) | | Unet-Segmentation | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 161.589 ms | 0 - 83 MB | NPU | [Unet-Segmentation.onnx](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.onnx) | | Unet-Segmentation | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 115.373 ms | 4 - 136 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.tflite) | | Unet-Segmentation | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 113.677 ms | 9 - 118 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.dlc) | | Unet-Segmentation | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 119.883 ms | 17 - 120 MB | NPU | [Unet-Segmentation.onnx](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.onnx) | | Unet-Segmentation | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 107.435 ms | 5 - 104 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.tflite) | | Unet-Segmentation | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 93.485 ms | 18 - 132 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.dlc) | | Unet-Segmentation | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 95.196 ms | 22 - 132 MB | NPU | [Unet-Segmentation.onnx](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.onnx) | | Unet-Segmentation | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 145.843 ms | 83 - 83 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.dlc) | | Unet-Segmentation | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 151.335 ms | 53 - 53 MB | NPU | [Unet-Segmentation.onnx](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation.onnx) | | Unet-Segmentation | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 127.6 ms | 2 - 48 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.tflite) | | Unet-Segmentation | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 1393.091 ms | 2 - 62 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.dlc) | | Unet-Segmentation | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 53.379 ms | 2 - 90 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.tflite) | | Unet-Segmentation | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 63.149 ms | 2 - 104 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.dlc) | | Unet-Segmentation | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 38.802 ms | 2 - 863 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.tflite) | | Unet-Segmentation | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 582.8 ms | 3 - 27 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.dlc) | | Unet-Segmentation | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 35.429 ms | 1 - 48 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.tflite) | | Unet-Segmentation | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 679.712 ms | 2 - 62 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.dlc) | | Unet-Segmentation | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | TFLITE | 308.118 ms | 2 - 262 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.tflite) | | Unet-Segmentation | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 331.8 ms | 2 - 268 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.dlc) | | Unet-Segmentation | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | TFLITE | 3127.378 ms | 0 - 846 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.tflite) | | Unet-Segmentation | w8a8 | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 127.6 ms | 2 - 48 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.tflite) | | Unet-Segmentation | w8a8 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 1393.091 ms | 2 - 62 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.dlc) | | Unet-Segmentation | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 38.641 ms | 0 - 885 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.tflite) | | Unet-Segmentation | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 583.129 ms | 5 - 23 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.dlc) | | Unet-Segmentation | w8a8 | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 68.159 ms | 2 - 49 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.tflite) | | Unet-Segmentation | w8a8 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 64.818 ms | 2 - 63 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.dlc) | | Unet-Segmentation | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 38.005 ms | 2 - 866 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.tflite) | | Unet-Segmentation | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 584.016 ms | 2 - 20 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.dlc) | | Unet-Segmentation | w8a8 | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 35.429 ms | 1 - 48 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.tflite) | | Unet-Segmentation | w8a8 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 679.712 ms | 2 - 62 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.dlc) | | Unet-Segmentation | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 39.081 ms | 0 - 869 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.tflite) | | Unet-Segmentation | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 582.829 ms | 2 - 21 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.dlc) | | Unet-Segmentation | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 131.566 ms | 0 - 200 MB | NPU | [Unet-Segmentation.onnx](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.onnx) | | Unet-Segmentation | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 28.523 ms | 1 - 88 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.tflite) | | Unet-Segmentation | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 449.848 ms | 2 - 97 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.dlc) | | Unet-Segmentation | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 110.057 ms | 3 - 676 MB | NPU | [Unet-Segmentation.onnx](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.onnx) | | Unet-Segmentation | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 26.66 ms | 1 - 51 MB | NPU | [Unet-Segmentation.tflite](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.tflite) | | Unet-Segmentation | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 423.193 ms | 2 - 70 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.dlc) | | Unet-Segmentation | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 92.43 ms | 18 - 767 MB | NPU | [Unet-Segmentation.onnx](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.onnx) | | Unet-Segmentation | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 609.107 ms | 35 - 35 MB | NPU | [Unet-Segmentation.dlc](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.dlc) | | Unet-Segmentation | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 155.892 ms | 60 - 60 MB | NPU | [Unet-Segmentation.onnx](https://huggingface.co/qualcomm/Unet-Segmentation/blob/main/Unet-Segmentation_w8a8.onnx) | ## Installation Install the package via pip: ```bash pip install qai-hub-models ``` ## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device Sign-in to [Qualcomm® AI Hub](https://app.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://app.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.unet_segmentation.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.unet_segmentation.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.unet_segmentation.export ``` ``` Profiling Results ------------------------------------------------------------ Unet-Segmentation Device : cs_8275 (ANDROID 14) Runtime : TFLITE Estimated inference time (ms) : 953.0 Estimated peak memory usage (MB): [0, 98] Total # Ops : 32 Compute Unit(s) : npu (32 ops) gpu (0 ops) cpu (0 ops) ``` ## How does this work? This [export script](https://aihub.qualcomm.com/models/unet_segmentation/qai_hub_models/models/Unet-Segmentation/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.unet_segmentation import Model # Load the model torch_model = Model.from_pretrained() # Device device = hub.Device("Samsung Galaxy S24") # 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. [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.unet_segmentation.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.unet_segmentation.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 Unet-Segmentation's performance across various devices [here](https://aihub.qualcomm.com/models/unet_segmentation). Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/) ## License * The license for the original implementation of Unet-Segmentation can be found [here](https://github.com/milesial/Pytorch-UNet/blob/master/LICENSE). * The license for the compiled assets for on-device deployment can be found [here](https://github.com/milesial/Pytorch-UNet/blob/master/LICENSE) ## References * [U-Net: Convolutional Networks for Biomedical Image Segmentation](https://arxiv.org/abs/1505.04597) * [Source Model Implementation](https://github.com/milesial/Pytorch-UNet) ## 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).