Hugging Face's logo

qualcomm
/
PSPNet

Image Segmentation
PyTorch
bu_auto
android
Model card Files
xet
 Community

PSPNet: Optimized for Qualcomm Devices

PSPNet (Pyramid Scene Parsing Network) is a semantic segmentation model that captures global context information by applying pyramid pooling modules. It is designed to improve scene understanding by aggregating contextual features at multiple scales.

This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the Qualcomm® AI Hub Models library to export with custom configurations. More details on model performance across various devices, can be found here.

Qualcomm AI Hub Models uses Qualcomm AI Hub Workbench to compile, profile, and evaluate this model. Sign up 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 1.25.0 Download
ONNX w8a8 Universal QAIRT 2.45, ONNX Runtime 1.25.0 Download
QNN_DLC float Universal QAIRT 2.45 Download
QNN_DLC w8a8 Universal QAIRT 2.45 Download
TFLITE float Universal QAIRT 2.45 Download
TFLITE w8a8 Universal QAIRT 2.45 Download

For more device-specific assets and performance metrics, visit PSPNet on Qualcomm® AI Hub.

Option 2: Export with Custom Configurations

Use the Qualcomm® AI Hub Models 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 PSPNet on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.semantic_segmentation

Model Stats:

  • Model checkpoint: pspnet101_ade20k.pth
  • Input resolution: 1x3x473x473
  • Number of parameters: 65.7M
  • Model size (float): 251 MB

Performance Summary

Model Runtime Precision Chipset Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit
PSPNet ONNX float Snapdragon® X2 Elite 428.778 ms 532 - 532 MB NPU
PSPNet ONNX float Snapdragon® X Elite 693.19 ms 528 - 528 MB NPU
PSPNet ONNX float Snapdragon® 8 Gen 3 Mobile 507.357 ms 15 - 1000 MB NPU
PSPNet ONNX float Snapdragon® 8 Gen 1 Mobile 1705.843 ms 1 - 449 MB NPU
PSPNet ONNX float Qualcomm® Dragonwing™ QCS8550 (Proxy) 591.723 ms 160 - 163 MB NPU
PSPNet ONNX float Qualcomm® QCS8450 1705.843 ms 1 - 449 MB NPU
PSPNet ONNX float Snapdragon® 8 Elite Gen 5 Mobile 321.374 ms 262 - 1049 MB NPU
PSPNet ONNX float Qualcomm® Dragonwing™ IQ-9075 1764.712 ms 11 - 56 MB NPU
PSPNet ONNX float Snapdragon® 8 Elite Mobile 337.626 ms 271 - 1030 MB NPU
PSPNet ONNX float Qualcomm® Dragonwing™ Q-8750 337.626 ms 271 - 1030 MB NPU
PSPNet ONNX float Qualcomm® Dragonwing™ IQ-X7181 693.19 ms 528 - 528 MB NPU
PSPNet ONNX w8a8 Snapdragon® X2 Elite 14.831 ms 133 - 133 MB NPU
PSPNet ONNX w8a8 Snapdragon® X Elite 27.702 ms 132 - 132 MB NPU
PSPNet ONNX w8a8 Snapdragon® 8 Gen 3 Mobile 22.249 ms 31 - 445 MB NPU
PSPNet ONNX w8a8 Snapdragon® 8 Gen 1 Mobile 41.042 ms 26 - 431 MB NPU
PSPNet ONNX w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 29.888 ms 69 - 72 MB NPU
PSPNet ONNX w8a8 Qualcomm® QCS8450 41.042 ms 26 - 431 MB NPU
PSPNet ONNX w8a8 Snapdragon® 8 Elite Mobile 18.766 ms 70 - 341 MB NPU
PSPNet ONNX w8a8 Snapdragon® 7 Gen 4 Mobile 93.034 ms 73 - 605 MB NPU
PSPNet ONNX w8a8 Snapdragon® 8 Elite Gen 5 Mobile 16.687 ms 69 - 413 MB NPU
PSPNet ONNX w8a8 Qualcomm® Dragonwing™ Q-7790 93.034 ms 73 - 605 MB NPU
PSPNet ONNX w8a8 Qualcomm® Dragonwing™ Q-8750 18.766 ms 70 - 341 MB NPU
PSPNet ONNX w8a8 Qualcomm® Dragonwing™ IQ-X7181 27.702 ms 132 - 132 MB NPU
PSPNet ONNX w8a8 Qualcomm® Dragonwing™ IQ-9075 33.674 ms 73 - 118 MB NPU
PSPNet QNN_DLC float Snapdragon® X2 Elite 228.176 ms 3 - 3 MB NPU
PSPNet QNN_DLC float Snapdragon® X Elite 532.47 ms 3 - 3 MB NPU
PSPNet QNN_DLC float Snapdragon® 8 Gen 3 Mobile 401.396 ms 1 - 950 MB NPU
PSPNet QNN_DLC float Snapdragon® 8 Gen 1 Mobile 1703.918 ms 0 - 438 MB NPU
PSPNet QNN_DLC float Qualcomm® QCS8275 1330.552 ms 2 - 727 MB NPU
PSPNet QNN_DLC float Qualcomm® Dragonwing™ QCS8550 (Proxy) 593.111 ms 3 - 5 MB NPU
PSPNet QNN_DLC float Qualcomm® QCS8450 1703.918 ms 0 - 438 MB NPU
PSPNet QNN_DLC float Snapdragon® 8 Elite Gen 5 Mobile 267.5 ms 3 - 757 MB NPU
PSPNet QNN_DLC float Qualcomm® Dragonwing™ IQ-9075 1762.009 ms 3 - 135 MB NPU
PSPNet QNN_DLC float Qualcomm® SA7255P 1330.552 ms 2 - 727 MB NPU
PSPNet QNN_DLC float Snapdragon® 8 Elite Mobile 302.172 ms 0 - 724 MB NPU
PSPNet QNN_DLC float Qualcomm® SA8295P 701.273 ms 3 - 329 MB NPU
PSPNet QNN_DLC float Qualcomm® Dragonwing™ Q-8750 302.172 ms 0 - 724 MB NPU
PSPNet QNN_DLC float Qualcomm® Dragonwing™ IQ-X7181 532.47 ms 3 - 3 MB NPU
PSPNet QNN_DLC w8a8 Snapdragon® X2 Elite 14.561 ms 1 - 1 MB NPU
PSPNet QNN_DLC w8a8 Snapdragon® X Elite 26.381 ms 1 - 1 MB NPU
PSPNet QNN_DLC w8a8 Snapdragon® 8 Gen 3 Mobile 20.421 ms 1 - 354 MB NPU
PSPNet QNN_DLC w8a8 Snapdragon® 8 Gen 1 Mobile 37.647 ms 1 - 353 MB NPU
PSPNet QNN_DLC w8a8 Qualcomm® QCS8275 63.865 ms 1 - 259 MB NPU
PSPNet QNN_DLC w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 27.107 ms 1 - 8 MB NPU
PSPNet QNN_DLC w8a8 Qualcomm® SA8775P 28.032 ms 1 - 261 MB NPU
PSPNet QNN_DLC w8a8 Qualcomm® SA8650P 28.032 ms 1 - 261 MB NPU
PSPNet QNN_DLC w8a8 Qualcomm® SA8255P 28.032 ms 1 - 261 MB NPU
PSPNet QNN_DLC w8a8 Qualcomm® QCS8450 37.647 ms 1 - 353 MB NPU
PSPNet QNN_DLC w8a8 Snapdragon® 8 Elite Mobile 17.906 ms 1 - 231 MB NPU
PSPNet QNN_DLC w8a8 Snapdragon® 7 Gen 4 Mobile 90.98 ms 1 - 438 MB NPU
PSPNet QNN_DLC w8a8 Snapdragon® 8 Elite Gen 5 Mobile 15.201 ms 1 - 312 MB NPU
PSPNet QNN_DLC w8a8 Qualcomm® SA7255P 63.865 ms 1 - 259 MB NPU
PSPNet QNN_DLC w8a8 Qualcomm® SA8295P 38.471 ms 1 - 263 MB NPU
PSPNet QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-7790 90.98 ms 1 - 438 MB NPU
PSPNet QNN_DLC w8a8 Qualcomm® Dragonwing™ Q-8750 17.906 ms 1 - 231 MB NPU
PSPNet QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-X7181 26.381 ms 1 - 1 MB NPU
PSPNet QNN_DLC w8a8 Qualcomm® Dragonwing™ IQ-9075 31.773 ms 1 - 35 MB NPU
PSPNet TFLITE float Snapdragon® 8 Gen 3 Mobile 405.108 ms 119 - 1113 MB NPU
PSPNet TFLITE float Snapdragon® 8 Gen 1 Mobile 1505.632 ms 14 - 563 MB NPU
PSPNet TFLITE float Qualcomm® QCS8275 1332.489 ms 128 - 901 MB NPU
PSPNet TFLITE float Qualcomm® Dragonwing™ QCS8550 (Proxy) 594.916 ms 128 - 132 MB NPU
PSPNet TFLITE float Qualcomm® SA8775P 649.209 ms 128 - 901 MB NPU
PSPNet TFLITE float Qualcomm® SA8650P 649.209 ms 128 - 901 MB NPU
PSPNet TFLITE float Qualcomm® SA8255P 649.209 ms 128 - 901 MB NPU
PSPNet TFLITE float Qualcomm® QCS8450 1505.632 ms 14 - 563 MB NPU
PSPNet TFLITE float Snapdragon® 8 Elite Gen 5 Mobile 269.126 ms 110 - 911 MB NPU
PSPNet TFLITE float Qualcomm® Dragonwing™ IQ-9075 646.334 ms 74 - 342 MB NPU
PSPNet TFLITE float Qualcomm® SA7255P 1332.489 ms 128 - 901 MB NPU
PSPNet TFLITE float Snapdragon® 8 Elite Mobile 302.274 ms 110 - 883 MB NPU
PSPNet TFLITE float Qualcomm® SA8295P 754.444 ms 128 - 556 MB NPU
PSPNet TFLITE float Qualcomm® Dragonwing™ Q-8750 302.274 ms 110 - 883 MB NPU
PSPNet TFLITE w8a8 Snapdragon® 8 Gen 3 Mobile 20.505 ms 30 - 391 MB NPU
PSPNet TFLITE w8a8 Snapdragon® 8 Gen 1 Mobile 37.365 ms 30 - 385 MB NPU
PSPNet TFLITE w8a8 Qualcomm® QCS8275 63.972 ms 32 - 283 MB NPU
PSPNet TFLITE w8a8 Qualcomm® Dragonwing™ QCS8550 (Proxy) 26.771 ms 32 - 34 MB NPU
PSPNet TFLITE w8a8 Qualcomm® SA8775P 27.873 ms 32 - 283 MB NPU
PSPNet TFLITE w8a8 Qualcomm® SA8650P 27.873 ms 32 - 283 MB NPU
PSPNet TFLITE w8a8 Qualcomm® SA8255P 27.873 ms 32 - 283 MB NPU
PSPNet TFLITE w8a8 Qualcomm® QCS8450 37.365 ms 30 - 385 MB NPU
PSPNet TFLITE w8a8 Snapdragon® 8 Elite Mobile 19.215 ms 32 - 269 MB NPU
PSPNet TFLITE w8a8 Snapdragon® 7 Gen 4 Mobile 90.62 ms 32 - 447 MB NPU
PSPNet TFLITE w8a8 Snapdragon® 8 Elite Gen 5 Mobile 15.148 ms 32 - 318 MB NPU
PSPNet TFLITE w8a8 Qualcomm® SA7255P 63.972 ms 32 - 283 MB NPU
PSPNet TFLITE w8a8 Qualcomm® SA8295P 38.217 ms 23 - 276 MB NPU
PSPNet TFLITE w8a8 Qualcomm® Dragonwing™ Q-7790 90.62 ms 32 - 447 MB NPU
PSPNet TFLITE w8a8 Qualcomm® Dragonwing™ Q-8750 19.215 ms 32 - 269 MB NPU
PSPNet TFLITE w8a8 Qualcomm® Dragonwing™ IQ-9075 41.107 ms 32 - 132 MB NPU

License

  • The license for the original implementation of PSPNet can be found here.

Community

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
Image Segmentation
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for qualcomm/PSPNet

Finetunes
1 model