README.md

---
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/inception_v3/web-assets/model_demo.png)

# Inception-v3: Optimized for Qualcomm Devices

InceptionNetV3 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 is based on the implementation of Inception-v3 found [here](https://github.com/pytorch/vision/blob/main/torchvision/models/inception.py).
This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.59.0/src/qai_hub_models/models/inception_v3) 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
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 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/inception_v3/releases/v0.59.0/inception_v3-onnx-float.zip)
| ONNX | w8a8 | Universal | QAIRT 2.45, ONNX Runtime 1.27.1 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/inception_v3/releases/v0.59.0/inception_v3-onnx-w8a8.zip)
| QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/inception_v3/releases/v0.59.0/inception_v3-qnn_dlc-float.zip)
| QNN_DLC | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/inception_v3/releases/v0.59.0/inception_v3-qnn_dlc-w8a8.zip)
| TFLITE | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/inception_v3/releases/v0.59.0/inception_v3-tflite-float.zip)
| TFLITE | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/inception_v3/releases/v0.59.0/inception_v3-tflite-w8a8.zip)

For more device-specific assets and performance metrics, visit **[Inception-v3 on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/inception_v3)**.


### Option 2: Export with Custom Configurations

Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.59.0/src/qai_hub_models/models/inception_v3) 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 [Inception-v3 on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.59.0/src/qai_hub_models/models/inception_v3) for usage instructions.

## Model Details

**Model Type:** Model_use_case.image_classification

**Model Stats:**
- Model checkpoint: Imagenet
- Input resolution: 224x224
- Number of parameters: 23.9M
- Model size (float): 90.9 MB
- Model size (w8a8): 23.3 MB

## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| Inception-v3 | ONNX | float | Snapdragon® X2 Elite | 1.128 ms | 1 - 1 MB | NPU
| Inception-v3 | ONNX | float | Snapdragon® X Elite | 2.149 ms | 46 - 46 MB | NPU
| Inception-v3 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 1.593 ms | 0 - 64 MB | NPU
| Inception-v3 | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 3.943 ms | 0 - 65 MB | NPU
| Inception-v3 | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 2.186 ms | 0 - 49 MB | NPU
| Inception-v3 | ONNX | float | Qualcomm® QCS8450 | 3.943 ms | 0 - 65 MB | NPU
| Inception-v3 | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 3.662 ms | 1 - 5 MB | NPU
| Inception-v3 | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 2.149 ms | 46 - 46 MB | NPU
| Inception-v3 | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 1.342 ms | 0 - 39 MB | NPU
| Inception-v3 | ONNX | float | Snapdragon® 8 Elite Mobile | 1.342 ms | 0 - 39 MB | NPU
| Inception-v3 | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.088 ms | 0 - 40 MB | NPU
| Inception-v3 | ONNX | w8a8 | Snapdragon® X2 Elite | 0.372 ms | 1 - 1 MB | NPU
| Inception-v3 | ONNX | w8a8 | Snapdragon® X Elite | 0.797 ms | 24 - 24 MB | NPU
| Inception-v3 | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.633 ms | 0 - 101 MB | NPU
| Inception-v3 | ONNX | w8a8 | Snapdragon® 8 Gen 1 Mobile | 1.227 ms | 0 - 101 MB | NPU
| Inception-v3 | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 4.068 ms | 0 - 3 MB | NPU
| Inception-v3 | ONNX | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.848 ms | 0 - 26 MB | NPU
| Inception-v3 | ONNX | w8a8 | Qualcomm® QCS8450 | 1.227 ms | 0 - 101 MB | NPU
| Inception-v3 | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 0.932 ms | 0 - 3 MB | NPU
| Inception-v3 | ONNX | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 0.797 ms | 24 - 24 MB | NPU
| Inception-v3 | ONNX | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.537 ms | 0 - 64 MB | NPU
| Inception-v3 | ONNX | w8a8 | Snapdragon® 8 Elite Mobile | 0.537 ms | 0 - 64 MB | NPU
| Inception-v3 | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.426 ms | 0 - 61 MB | NPU
| Inception-v3 | QNN_DLC | float | Snapdragon® X2 Elite | 1.189 ms | 1 - 1 MB | NPU
| Inception-v3 | QNN_DLC | float | Snapdragon® X Elite | 2.093 ms | 1 - 1 MB | NPU
| Inception-v3 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 1.51 ms | 0 - 59 MB | NPU
| Inception-v3 | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 3.73 ms | 0 - 60 MB | NPU
| Inception-v3 | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8275 | 13.486 ms | 1 - 35 MB | NPU
| Inception-v3 | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 1.98 ms | 0 - 9 MB | NPU
| Inception-v3 | QNN_DLC | float | Qualcomm® SA8775P | 3.595 ms | 1 - 40 MB | NPU
| Inception-v3 | QNN_DLC | float | Qualcomm® SA8650P | 3.595 ms | 1 - 40 MB | NPU
| Inception-v3 | QNN_DLC | float | Qualcomm® SA8255P | 3.595 ms | 1 - 40 MB | NPU
| Inception-v3 | QNN_DLC | float | Qualcomm® QCS8450 | 3.73 ms | 0 - 60 MB | NPU
| Inception-v3 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 3.422 ms | 3 - 6 MB | NPU
| Inception-v3 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 2.093 ms | 1 - 1 MB | NPU
| Inception-v3 | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 1.261 ms | 0 - 36 MB | NPU
| Inception-v3 | QNN_DLC | float | Qualcomm® SA7255P | 13.486 ms | 1 - 35 MB | NPU
| Inception-v3 | QNN_DLC | float | Qualcomm® SA8295P | 4.182 ms | 1 - 34 MB | NPU
| Inception-v3 | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 1.261 ms | 0 - 36 MB | NPU
| Inception-v3 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.017 ms | 1 - 41 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 0.459 ms | 0 - 0 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Snapdragon® X Elite | 0.862 ms | 0 - 0 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.592 ms | 0 - 84 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Snapdragon® 8 Gen 1 Mobile | 1.186 ms | 0 - 83 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 4.059 ms | 2 - 4 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS8275 | 2.39 ms | 0 - 54 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.807 ms | 0 - 2 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® SA8775P | 0.996 ms | 0 - 55 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® SA8650P | 0.996 ms | 0 - 55 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® SA8255P | 0.996 ms | 0 - 55 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® QCS8450 | 1.186 ms | 0 - 83 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 0.939 ms | 0 - 2 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ IQ-X7181 | 0.862 ms | 0 - 0 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 10.573 ms | 0 - 168 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 1.308 ms | 0 - 55 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.535 ms | 0 - 58 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® SA7255P | 2.39 ms | 0 - 54 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Qualcomm® SA8295P | 1.594 ms | 0 - 52 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Snapdragon® 8 Elite Mobile | 0.535 ms | 0 - 58 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.412 ms | 0 - 53 MB | NPU
| Inception-v3 | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 1.308 ms | 0 - 55 MB | NPU
| Inception-v3 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 1.483 ms | 0 - 116 MB | NPU
| Inception-v3 | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 3.733 ms | 0 - 122 MB | NPU
| Inception-v3 | TFLITE | float | Qualcomm® Dragonwing™ QCS8275 | 13.505 ms | 0 - 66 MB | NPU
| Inception-v3 | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 2.027 ms | 0 - 2 MB | NPU
| Inception-v3 | TFLITE | float | Qualcomm® SA8775P | 3.611 ms | 0 - 69 MB | NPU
| Inception-v3 | TFLITE | float | Qualcomm® SA8650P | 3.611 ms | 0 - 69 MB | NPU
| Inception-v3 | TFLITE | float | Qualcomm® SA8255P | 3.611 ms | 0 - 69 MB | NPU
| Inception-v3 | TFLITE | float | Qualcomm® QCS8450 | 3.733 ms | 0 - 122 MB | NPU
| Inception-v3 | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 3.426 ms | 0 - 49 MB | NPU
| Inception-v3 | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 1.276 ms | 0 - 68 MB | NPU
| Inception-v3 | TFLITE | float | Qualcomm® SA7255P | 13.505 ms | 0 - 66 MB | NPU
| Inception-v3 | TFLITE | float | Qualcomm® SA8295P | 4.155 ms | 0 - 64 MB | NPU
| Inception-v3 | TFLITE | float | Snapdragon® 8 Elite Mobile | 1.276 ms | 0 - 68 MB | NPU
| Inception-v3 | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 1.011 ms | 0 - 69 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 0.639 ms | 0 - 81 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Snapdragon® 8 Gen 1 Mobile | 1.287 ms | 0 - 86 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS6490 | 4.235 ms | 0 - 25 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS8275 | 2.551 ms | 0 - 51 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 0.909 ms | 0 - 33 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Qualcomm® SA8775P | 1.181 ms | 0 - 55 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Qualcomm® SA8650P | 1.181 ms | 0 - 55 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Qualcomm® SA8255P | 1.181 ms | 0 - 55 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Qualcomm® QCS8450 | 1.287 ms | 0 - 86 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Qualcomm® Dragonwing™ IQ-9075 | 1.046 ms | 0 - 25 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-6690 | 10.657 ms | 0 - 168 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-7790 | 1.403 ms | 0 - 53 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Qualcomm® Dragonwing™ Q-8750 | 0.553 ms | 0 - 57 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Qualcomm® SA7255P | 2.551 ms | 0 - 51 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Qualcomm® SA8295P | 1.809 ms | 0 - 50 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Snapdragon® 8 Elite Mobile | 0.553 ms | 0 - 57 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.455 ms | 0 - 52 MB | NPU
| Inception-v3 | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 1.403 ms | 0 - 53 MB | NPU

## License
* The license for the original implementation of Inception-v3 can be found
  [here](https://github.com/pytorch/vision/blob/main/LICENSE).

## References
* [Rethinking the Inception Architecture for Computer Vision](http://arxiv.org/abs/1512.00567)
* [Source Model Implementation](https://github.com/pytorch/vision/blob/main/torchvision/models/inception.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).