--- library_name: pytorch license: other tags: - android pipeline_tag: image-segmentation --- ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/ffnet_122ns_lowres/web-assets/model_demo.png) # FFNet-122NS-LowRes: Optimized for Mobile Deployment ## Semantic segmentation for automotive street scenes FFNet-122NS-LowRes is a "fuss-free network" that segments street scene images with per-pixel classes like road, sidewalk, and pedestrian. Trained on the Cityscapes dataset. This model is an implementation of FFNet-122NS-LowRes found [here](https://github.com/Qualcomm-AI-research/FFNet). This repository provides scripts to run FFNet-122NS-LowRes on Qualcomm® devices. More details on model performance across various devices, can be found [here](https://aihub.qualcomm.com/models/ffnet_122ns_lowres). ### Model Details - **Model Type:** Model_use_case.semantic_segmentation - **Model Stats:** - Model checkpoint: ffnet122NS_CCC_cityscapes_state_dict_quarts_pre_down - Input resolution: 1024x512 - Number of output classes: 19 - Number of parameters: 32.1M - Model size (float): 123 MB - Model size (w8a8): 31.3 MB | Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model |---|---|---|---|---|---|---|---|---| | FFNet-122NS-LowRes | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 37.296 ms | 1 - 167 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.tflite) | | FFNet-122NS-LowRes | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 38.019 ms | 5 - 141 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.dlc) | | FFNet-122NS-LowRes | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 23.798 ms | 1 - 245 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.tflite) | | FFNet-122NS-LowRes | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 23.717 ms | 6 - 180 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.dlc) | | FFNet-122NS-LowRes | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 11.093 ms | 1 - 3 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.tflite) | | FFNet-122NS-LowRes | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 11.14 ms | 6 - 8 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.dlc) | | FFNet-122NS-LowRes | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 7.064 ms | 0 - 60 MB | NPU | [FFNet-122NS-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.onnx.zip) | | FFNet-122NS-LowRes | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 14.43 ms | 1 - 167 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.tflite) | | FFNet-122NS-LowRes | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 14.286 ms | 1 - 137 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.dlc) | | FFNet-122NS-LowRes | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 37.296 ms | 1 - 167 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.tflite) | | FFNet-122NS-LowRes | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 38.019 ms | 5 - 141 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.dlc) | | FFNet-122NS-LowRes | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 15.547 ms | 1 - 166 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.tflite) | | FFNet-122NS-LowRes | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 16.461 ms | 0 - 134 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.dlc) | | FFNet-122NS-LowRes | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 14.43 ms | 1 - 167 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.tflite) | | FFNet-122NS-LowRes | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 14.286 ms | 1 - 137 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.dlc) | | FFNet-122NS-LowRes | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 7.406 ms | 0 - 252 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.tflite) | | FFNet-122NS-LowRes | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 7.712 ms | 6 - 181 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.dlc) | | FFNet-122NS-LowRes | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 5.028 ms | 7 - 168 MB | NPU | [FFNet-122NS-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.onnx.zip) | | FFNet-122NS-LowRes | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 5.967 ms | 1 - 168 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.tflite) | | FFNet-122NS-LowRes | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 5.746 ms | 6 - 146 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.dlc) | | FFNet-122NS-LowRes | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 6.216 ms | 2 - 124 MB | NPU | [FFNet-122NS-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.onnx.zip) | | FFNet-122NS-LowRes | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | TFLITE | 5.092 ms | 1 - 172 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.tflite) | | FFNet-122NS-LowRes | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | QNN_DLC | 5.772 ms | 6 - 150 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.dlc) | | FFNet-122NS-LowRes | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | ONNX | 3.71 ms | 4 - 128 MB | NPU | [FFNet-122NS-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.onnx.zip) | | FFNet-122NS-LowRes | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 11.787 ms | 6 - 6 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.dlc) | | FFNet-122NS-LowRes | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 7.052 ms | 56 - 56 MB | NPU | [FFNet-122NS-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes.onnx.zip) | | FFNet-122NS-LowRes | w8a8 | Dragonwing Q-6690 MTP | Qualcomm® QCM6690 | TFLITE | 22.263 ms | 0 - 172 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.tflite) | | FFNet-122NS-LowRes | w8a8 | Dragonwing Q-6690 MTP | Qualcomm® QCM6690 | QNN_DLC | 27.722 ms | 2 - 176 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.dlc) | | FFNet-122NS-LowRes | w8a8 | Dragonwing Q-6690 MTP | Qualcomm® QCM6690 | ONNX | 84.652 ms | 58 - 77 MB | CPU | [FFNet-122NS-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.onnx.zip) | | FFNet-122NS-LowRes | w8a8 | Dragonwing RB3 Gen 2 Vision Kit | Qualcomm® QCS6490 | TFLITE | 8.303 ms | 0 - 35 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.tflite) | | FFNet-122NS-LowRes | w8a8 | Dragonwing RB3 Gen 2 Vision Kit | Qualcomm® QCS6490 | QNN_DLC | 13.935 ms | 0 - 4 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.dlc) | | FFNet-122NS-LowRes | w8a8 | Dragonwing RB3 Gen 2 Vision Kit | Qualcomm® QCS6490 | ONNX | 91.893 ms | 54 - 155 MB | CPU | [FFNet-122NS-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.onnx.zip) | | FFNet-122NS-LowRes | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 5.957 ms | 0 - 164 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.tflite) | | FFNet-122NS-LowRes | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 9.278 ms | 2 - 169 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.dlc) | | FFNet-122NS-LowRes | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 3.278 ms | 0 - 232 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.tflite) | | FFNet-122NS-LowRes | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 6.874 ms | 2 - 225 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.dlc) | | FFNet-122NS-LowRes | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 2.592 ms | 0 - 3 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.tflite) | | FFNet-122NS-LowRes | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 4.466 ms | 2 - 4 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.dlc) | | FFNet-122NS-LowRes | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 2.848 ms | 0 - 33 MB | NPU | [FFNet-122NS-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.onnx.zip) | | FFNet-122NS-LowRes | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 3.077 ms | 0 - 164 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.tflite) | | FFNet-122NS-LowRes | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 21.231 ms | 2 - 168 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.dlc) | | FFNet-122NS-LowRes | w8a8 | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 5.957 ms | 0 - 164 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.tflite) | | FFNet-122NS-LowRes | w8a8 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 9.278 ms | 2 - 169 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.dlc) | | FFNet-122NS-LowRes | w8a8 | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 3.895 ms | 0 - 172 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.tflite) | | FFNet-122NS-LowRes | w8a8 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 6.037 ms | 2 - 175 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.dlc) | | FFNet-122NS-LowRes | w8a8 | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 3.077 ms | 0 - 164 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.tflite) | | FFNet-122NS-LowRes | w8a8 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 21.231 ms | 2 - 168 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.dlc) | | FFNet-122NS-LowRes | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 1.876 ms | 0 - 224 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.tflite) | | FFNet-122NS-LowRes | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 3.17 ms | 2 - 224 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.dlc) | | FFNet-122NS-LowRes | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 2.045 ms | 0 - 222 MB | NPU | [FFNet-122NS-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.onnx.zip) | | FFNet-122NS-LowRes | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 1.468 ms | 0 - 165 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.tflite) | | FFNet-122NS-LowRes | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 2.195 ms | 2 - 171 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.dlc) | | FFNet-122NS-LowRes | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 1.691 ms | 0 - 158 MB | NPU | [FFNet-122NS-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.onnx.zip) | | FFNet-122NS-LowRes | w8a8 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | TFLITE | 3.865 ms | 0 - 168 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.tflite) | | FFNet-122NS-LowRes | w8a8 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | QNN_DLC | 5.733 ms | 2 - 178 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.dlc) | | FFNet-122NS-LowRes | w8a8 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | ONNX | 83.947 ms | 58 - 79 MB | CPU | [FFNet-122NS-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.onnx.zip) | | FFNet-122NS-LowRes | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | TFLITE | 1.271 ms | 0 - 168 MB | NPU | [FFNet-122NS-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.tflite) | | FFNet-122NS-LowRes | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | QNN_DLC | 1.795 ms | 2 - 172 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.dlc) | | FFNet-122NS-LowRes | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | ONNX | 1.525 ms | 0 - 155 MB | NPU | [FFNet-122NS-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.onnx.zip) | | FFNet-122NS-LowRes | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 4.807 ms | 2 - 2 MB | NPU | [FFNet-122NS-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.dlc) | | FFNet-122NS-LowRes | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 2.754 ms | 30 - 30 MB | NPU | [FFNet-122NS-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-122NS-LowRes/blob/main/FFNet-122NS-LowRes_w8a8.onnx.zip) | ## Installation Install the package via pip: ```bash # NOTE: 3.10 <= PYTHON_VERSION < 3.14 is supported. pip install "qai-hub-models[ffnet-122ns-lowres]" ``` ## 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.ffnet_122ns_lowres.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.ffnet_122ns_lowres.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.ffnet_122ns_lowres.export ``` ## How does this work? This [export script](https://aihub.qualcomm.com/models/ffnet_122ns_lowres/qai_hub_models/models/FFNet-122NS-LowRes/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.ffnet_122ns_lowres 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.ffnet_122ns_lowres.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.ffnet_122ns_lowres.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 FFNet-122NS-LowRes's performance across various devices [here](https://aihub.qualcomm.com/models/ffnet_122ns_lowres). Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/) ## License * The license for the original implementation of FFNet-122NS-LowRes can be found [here](https://github.com/Qualcomm-AI-research/FFNet/blob/master/LICENSE). ## References * [Simple and Efficient Architectures for Semantic Segmentation](https://arxiv.org/abs/2206.08236) * [Source Model Implementation](https://github.com/Qualcomm-AI-research/FFNet) ## 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).