v0.46.1

See https://github.com/quic/ai-hub-models/releases/v0.46.1 for changelog.

README.md

@@ -10,261 +10,90 @@ pipeline_tag: image-segmentation
 
 ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/yolov8_seg/web-assets/model_demo.png)
 
-# YOLOv8-Segmentation: Optimized for Mobile Deployment
-## Real-time object segmentation optimized for mobile and edge by Ultralytics
-
+# YOLOv8-Segmentation: Optimized for Qualcomm Devices
 
 Ultralytics YOLOv8 is a machine learning model that predicts bounding boxes, segmentation masks and classes of objects in an image.
 
-This model is an implementation of YOLOv8-Segmentation found [here](https://github.com/ultralytics/ultralytics/tree/main/ultralytics/models/yolo/segment).
-
-
-This repository provides scripts to run YOLOv8-Segmentation on Qualcomm® devices.
-More details on model performance across various devices, can be found
-[here](https://aihub.qualcomm.com/models/yolov8_seg).
-
-**WARNING**: The model assets are not readily available for download due to licensing restrictions.
-
-### Model Details
-
-- **Model Type:** Model_use_case.semantic_segmentation
-- **Model Stats:**
-  - Model checkpoint: YOLOv8N-Seg
-  - Input resolution: 640x640
-  - Number of output classes: 80
-  - Number of parameters: 3.43M
-  - Model size (float): 13.2 MB
-  - Model size (w8a16): 3.91 MB
-
-| Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
-|---|---|---|---|---|---|---|---|---|
-| YOLOv8-Segmentation | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 16.32 ms | 4 - 220 MB | NPU | -- |
-| YOLOv8-Segmentation | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 16.257 ms | 0 - 238 MB | NPU | -- |
-| YOLOv8-Segmentation | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 8.246 ms | 4 - 183 MB | NPU | -- |
-| YOLOv8-Segmentation | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 8.234 ms | 5 - 182 MB | NPU | -- |
-| YOLOv8-Segmentation | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 4.05 ms | 4 - 14 MB | NPU | -- |
-| YOLOv8-Segmentation | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 4.05 ms | 5 - 7 MB | NPU | -- |
-| YOLOv8-Segmentation | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 5.909 ms | 9 - 17 MB | NPU | -- |
-| YOLOv8-Segmentation | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 6.041 ms | 4 - 203 MB | NPU | -- |
-| YOLOv8-Segmentation | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 5.918 ms | 1 - 202 MB | NPU | -- |
-| YOLOv8-Segmentation | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 16.32 ms | 4 - 220 MB | NPU | -- |
-| YOLOv8-Segmentation | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 16.257 ms | 0 - 238 MB | NPU | -- |
-| YOLOv8-Segmentation | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 8.593 ms | 4 - 164 MB | NPU | -- |
-| YOLOv8-Segmentation | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 8.421 ms | 2 - 154 MB | NPU | -- |
-| YOLOv8-Segmentation | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 6.041 ms | 4 - 203 MB | NPU | -- |
-| YOLOv8-Segmentation | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 5.918 ms | 1 - 202 MB | NPU | -- |
-| YOLOv8-Segmentation | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 2.991 ms | 0 - 378 MB | NPU | -- |
-| YOLOv8-Segmentation | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 2.994 ms | 5 - 383 MB | NPU | -- |
-| YOLOv8-Segmentation | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 3.992 ms | 2 - 218 MB | NPU | -- |
-| YOLOv8-Segmentation | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 2.362 ms | 0 - 208 MB | NPU | -- |
-| YOLOv8-Segmentation | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 2.288 ms | 5 - 218 MB | NPU | -- |
-| YOLOv8-Segmentation | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 3.362 ms | 1 - 172 MB | NPU | -- |
-| YOLOv8-Segmentation | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | TFLITE | 1.805 ms | 0 - 243 MB | NPU | -- |
-| YOLOv8-Segmentation | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | QNN_DLC | 1.809 ms | 5 - 237 MB | NPU | -- |
-| YOLOv8-Segmentation | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | ONNX | 2.808 ms | 2 - 157 MB | NPU | -- |
-| YOLOv8-Segmentation | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 4.47 ms | 5 - 5 MB | NPU | -- |
-| YOLOv8-Segmentation | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 6.024 ms | 17 - 17 MB | NPU | -- |
-| YOLOv8-Segmentation | w8a16 | Dragonwing Q-6690 MTP | Qualcomm® QCM6690 | QNN_DLC | 27.155 ms | 2 - 156 MB | NPU | -- |
-| YOLOv8-Segmentation | w8a16 | Dragonwing RB3 Gen 2 Vision Kit | Qualcomm® QCS6490 | QNN_DLC | 11.139 ms | 1 - 7 MB | NPU | -- |
-| YOLOv8-Segmentation | w8a16 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 7.679 ms | 1 - 149 MB | NPU | -- |
-| YOLOv8-Segmentation | w8a16 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 4.701 ms | 2 - 178 MB | NPU | -- |
-| YOLOv8-Segmentation | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 3.809 ms | 2 - 4 MB | NPU | -- |
-| YOLOv8-Segmentation | w8a16 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 18.034 ms | 1 - 151 MB | NPU | -- |
-| YOLOv8-Segmentation | w8a16 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 7.679 ms | 1 - 149 MB | NPU | -- |
-| YOLOv8-Segmentation | w8a16 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 5.129 ms | 0 - 154 MB | NPU | -- |
-| YOLOv8-Segmentation | w8a16 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 18.034 ms | 1 - 151 MB | NPU | -- |
-| YOLOv8-Segmentation | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 2.523 ms | 2 - 176 MB | NPU | -- |
-| YOLOv8-Segmentation | w8a16 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 1.697 ms | 2 - 162 MB | NPU | -- |
-| YOLOv8-Segmentation | w8a16 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | QNN_DLC | 4.784 ms | 2 - 158 MB | NPU | -- |
-| YOLOv8-Segmentation | w8a16 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen 5 Mobile | QNN_DLC | 1.428 ms | 2 - 157 MB | NPU | -- |
-| YOLOv8-Segmentation | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 4.253 ms | 2 - 2 MB | NPU | -- |
-
-
-
-
-## Installation
-
-
-Install the package via pip:
-```bash
-# NOTE: 3.10 <= PYTHON_VERSION < 3.14 is supported.
-pip install "qai-hub-models[yolov8-seg]"
-```
-
-
-## 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.yolov8_seg.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.yolov8_seg.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.yolov8_seg.export
-```
-
-
-
-## How does this work?
-
-This [export script](https://aihub.qualcomm.com/models/yolov8_seg/qai_hub_models/models/YOLOv8-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.yolov8_seg 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.yolov8_seg.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.yolov8_seg.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 YOLOv8-Segmentation's performance across various devices [here](https://aihub.qualcomm.com/models/yolov8_seg).
-Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
-
+This is based on the implementation of YOLOv8-Segmentation found [here](https://github.com/ultralytics/ultralytics/tree/main/ultralytics/models/yolo/segment).
+This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/yolov8_seg) 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.com) 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
+Due to licensing restrictions, we cannot distribute pre-exported model assets for this model.
+Use the [Qualcomm® AI Hub Models](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/yolov8_seg) 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
+
+See our repository for [YOLOv8-Segmentation on GitHub](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/yolov8_seg) for usage instructions.
+
+
+## Model Details
+
+**Model Type:** Model_use_case.semantic_segmentation
+
+**Model Stats:**
+- Model checkpoint: YOLOv8N-Seg
+- Input resolution: 640x640
+- Number of output classes: 80
+- Number of parameters: 3.43M
+- Model size (float): 13.2 MB
+- Model size (w8a16): 3.91 MB
+
+## Performance Summary
+| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
+|---|---|---|---|---|---|---
+| YOLOv8-Segmentation | ONNX | float | Snapdragon® X Elite | 6.013 ms | 17 - 17 MB | NPU
+| YOLOv8-Segmentation | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 3.989 ms | 17 - 226 MB | NPU
+| YOLOv8-Segmentation | ONNX | float | Qualcomm® QCS8550 (Proxy) | 5.897 ms | 0 - 43 MB | NPU
+| YOLOv8-Segmentation | ONNX | float | Qualcomm® QCS9075 | 8.08 ms | 13 - 16 MB | NPU
+| YOLOv8-Segmentation | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 3.285 ms | 0 - 155 MB | NPU
+| YOLOv8-Segmentation | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 2.759 ms | 0 - 156 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | float | Snapdragon® X Elite | 4.912 ms | 5 - 5 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 3.339 ms | 5 - 261 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 16.912 ms | 1 - 200 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 4.548 ms | 5 - 7 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | float | Qualcomm® SA8775P | 6.278 ms | 1 - 203 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | float | Qualcomm® QCS9075 | 6.029 ms | 5 - 15 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 9.835 ms | 5 - 199 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | float | Qualcomm® SA7255P | 16.912 ms | 1 - 200 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | float | Qualcomm® SA8295P | 9.2 ms | 2 - 166 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 2.648 ms | 0 - 198 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.927 ms | 5 - 203 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | w8a16 | Snapdragon® X Elite | 4.296 ms | 2 - 2 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | w8a16 | Snapdragon® 8 Gen 3 Mobile | 2.574 ms | 2 - 84 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | w8a16 | Qualcomm® QCS6490 | 11.362 ms | 3 - 9 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | w8a16 | Qualcomm® QCS8275 (Proxy) | 7.868 ms | 0 - 57 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | w8a16 | Qualcomm® QCS8550 (Proxy) | 3.839 ms | 2 - 4 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | w8a16 | Qualcomm® SA8775P | 4.516 ms | 0 - 61 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | w8a16 | Qualcomm® QCS9075 | 4.414 ms | 1 - 7 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | w8a16 | Qualcomm® QCM6690 | 27.628 ms | 2 - 179 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | w8a16 | Qualcomm® QCS8450 (Proxy) | 4.796 ms | 2 - 82 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | w8a16 | Qualcomm® SA7255P | 7.868 ms | 0 - 57 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | w8a16 | Qualcomm® SA8295P | 5.183 ms | 1 - 57 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | w8a16 | Snapdragon® 8 Elite For Galaxy Mobile | 1.787 ms | 2 - 64 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | w8a16 | Snapdragon® 7 Gen 4 Mobile | 4.64 ms | 2 - 178 MB | NPU
+| YOLOv8-Segmentation | QNN_DLC | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 1.416 ms | 2 - 65 MB | NPU
+| YOLOv8-Segmentation | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 2.941 ms | 0 - 176 MB | NPU
+| YOLOv8-Segmentation | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 16.176 ms | 4 - 107 MB | NPU
+| YOLOv8-Segmentation | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 3.981 ms | 0 - 2 MB | NPU
+| YOLOv8-Segmentation | TFLITE | float | Qualcomm® SA8775P | 23.252 ms | 4 - 108 MB | NPU
+| YOLOv8-Segmentation | TFLITE | float | Qualcomm® QCS9075 | 5.765 ms | 4 - 23 MB | NPU
+| YOLOv8-Segmentation | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 8.896 ms | 4 - 204 MB | NPU
+| YOLOv8-Segmentation | TFLITE | float | Qualcomm® SA7255P | 16.176 ms | 4 - 107 MB | NPU
+| YOLOv8-Segmentation | TFLITE | float | Qualcomm® SA8295P | 8.553 ms | 4 - 174 MB | NPU
+| YOLOv8-Segmentation | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 2.227 ms | 0 - 111 MB | NPU
+| YOLOv8-Segmentation | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.772 ms | 0 - 127 MB | NPU
 
 ## License
 * The license for the original implementation of YOLOv8-Segmentation can be found
   [here](https://github.com/ultralytics/ultralytics/blob/main/LICENSE).
 
-
-
 ## References
 * [Ultralytics YOLOv8 Docs: Instance Segmentation](https://docs.ultralytics.com/tasks/segment/)
 * [Source Model Implementation](https://github.com/ultralytics/ultralytics/tree/main/ultralytics/models/yolo/segment)
 
-
-
 ## 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).
-
-