---
library_name: pytorch
license: other
tags:
- backbone
- bu_auto
- android
pipeline_tag: image-classification

---

![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/web-assets/model_demo.png)

# EfficientNet-B4: Optimized for Qualcomm Devices

EfficientNetB4 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 EfficientNet-B4 found [here](https://github.com/pytorch/vision/blob/main/torchvision/models/efficientnet.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/v0.58.0/src/qai_hub_models/models/efficientnet_b4) 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 1.25.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/releases/v0.58.0/efficientnet_b4-onnx-float.zip)
| ONNX | w8a16 | Universal | QAIRT 2.45, ONNX Runtime 1.25.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/releases/v0.58.0/efficientnet_b4-onnx-w8a16.zip)
| QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/releases/v0.58.0/efficientnet_b4-qnn_dlc-float.zip)
| QNN_DLC | w8a16 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/releases/v0.58.0/efficientnet_b4-qnn_dlc-w8a16.zip)
| TFLITE | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/efficientnet_b4/releases/v0.58.0/efficientnet_b4-tflite-float.zip)

For more device-specific assets and performance metrics, visit **[EfficientNet-B4 on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/efficientnet_b4)**.


### Option 2: Export with Custom Configurations

Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.58.0/src/qai_hub_models/models/efficientnet_b4) 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 [EfficientNet-B4 on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.58.0/src/qai_hub_models/models/efficientnet_b4) for usage instructions.

## Model Details

**Model Type:** Model_use_case.image_classification

**Model Stats:**
- Model checkpoint: Imagenet
- Input resolution: 380x380
- Number of parameters: 19.3M
- Model size (float): 73.6 MB
- Model size (w8a16): 24.0 MB

## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| EfficientNet-B4 | ONNX | float | Snapdragon® X2 Elite | 3.918 ms | 2 - 2 MB | NPU
| EfficientNet-B4 | ONNX | float | Snapdragon® X Elite | 7.732 ms | 44 - 44 MB | NPU
| EfficientNet-B4 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 5.309 ms | 0 - 147 MB | NPU
| EfficientNet-B4 | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 20.415 ms | 0 - 191 MB | NPU
| EfficientNet-B4 | ONNX | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 7.322 ms | 0 - 49 MB | NPU
| EfficientNet-B4 | ONNX | float | Qualcomm® QCS8450 | 20.415 ms | 0 - 191 MB | NPU
| EfficientNet-B4 | ONNX | float | Qualcomm® Dragonwing™ IQ-9075 | 9.332 ms | 1 - 6 MB | NPU
| EfficientNet-B4 | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 3.114 ms | 0 - 206 MB | NPU
| EfficientNet-B4 | ONNX | float | Snapdragon® 8 Elite Mobile | 4.047 ms | 0 - 88 MB | NPU
| EfficientNet-B4 | ONNX | float | Qualcomm® Dragonwing™ Q-8750 | 4.047 ms | 0 - 88 MB | NPU
| EfficientNet-B4 | ONNX | float | Qualcomm® Dragonwing™ IQ-X7181 | 7.732 ms | 44 - 44 MB | NPU
| EfficientNet-B4 | ONNX | w8a16 | Snapdragon® X2 Elite | 2.951 ms | 2 - 2 MB | NPU
| EfficientNet-B4 | ONNX | w8a16 | Snapdragon® X Elite | 8.021 ms | 24 - 24 MB | NPU
| EfficientNet-B4 | ONNX | w8a16 | Snapdragon® 8 Gen 3 Mobile | 5.095 ms | 1 - 226 MB | NPU
| EfficientNet-B4 | ONNX | w8a16 | Snapdragon® 8 Gen 1 Mobile | 8.918 ms | 0 - 229 MB | NPU
| EfficientNet-B4 | ONNX | w8a16 | Qualcomm® Dragonwing™ QCS6490 | 32.645 ms | 0 - 4 MB | NPU
| EfficientNet-B4 | ONNX | w8a16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 7.637 ms | 0 - 30 MB | NPU
| EfficientNet-B4 | ONNX | w8a16 | Qualcomm® QCS8450 | 8.918 ms | 0 - 229 MB | NPU
| EfficientNet-B4 | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-9075 | 7.997 ms | 1 - 4 MB | NPU
| EfficientNet-B4 | ONNX | w8a16 | Snapdragon® 7 Gen 4 Mobile | 9.113 ms | 1 - 289 MB | NPU
| EfficientNet-B4 | ONNX | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 2.771 ms | 0 - 176 MB | NPU
| EfficientNet-B4 | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-6690 | 49.373 ms | 1 - 297 MB | NPU
| EfficientNet-B4 | ONNX | w8a16 | Snapdragon® 8 Elite Mobile | 3.457 ms | 0 - 167 MB | NPU
| EfficientNet-B4 | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-7790 | 9.113 ms | 1 - 289 MB | NPU
| EfficientNet-B4 | ONNX | w8a16 | Qualcomm® Dragonwing™ Q-8750 | 3.457 ms | 0 - 167 MB | NPU
| EfficientNet-B4 | ONNX | w8a16 | Qualcomm® Dragonwing™ IQ-X7181 | 8.021 ms | 24 - 24 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Snapdragon® X2 Elite | 4.563 ms | 2 - 2 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Snapdragon® X Elite | 8.937 ms | 2 - 2 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 5.802 ms | 0 - 145 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 22.973 ms | 2 - 187 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Qualcomm® QCS8275 | 29.123 ms | 2 - 81 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 8.119 ms | 2 - 162 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Qualcomm® SA8775P | 10.311 ms | 2 - 85 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Qualcomm® SA8650P | 10.311 ms | 2 - 85 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Qualcomm® SA8255P | 10.311 ms | 2 - 85 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Qualcomm® QCS8450 | 22.973 ms | 2 - 187 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-9075 | 10.019 ms | 4 - 7 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 3.351 ms | 2 - 208 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Qualcomm® SA7255P | 29.123 ms | 2 - 81 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 4.301 ms | 0 - 87 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Qualcomm® SA8295P | 18.789 ms | 2 - 124 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Qualcomm® Dragonwing™ Q-8750 | 4.301 ms | 0 - 87 MB | NPU
| EfficientNet-B4 | QNN_DLC | float | Qualcomm® Dragonwing™ IQ-X7181 | 8.937 ms | 2 - 2 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® X2 Elite | 3.556 ms | 1 - 1 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® X Elite | 9.14 ms | 1 - 1 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® 8 Gen 3 Mobile | 5.659 ms | 1 - 199 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® 8 Gen 1 Mobile | 11.438 ms | 1 - 202 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ QCS6490 | 33.289 ms | 3 - 5 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® QCS8275 | 15.489 ms | 1 - 142 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 8.387 ms | 1 - 3 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® SA8775P | 8.975 ms | 1 - 144 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® SA8650P | 8.975 ms | 1 - 144 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® SA8255P | 8.975 ms | 1 - 144 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® QCS8450 | 11.438 ms | 1 - 202 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-9075 | 8.699 ms | 3 - 5 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® 7 Gen 4 Mobile | 9.751 ms | 1 - 269 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® 8 Elite Gen 5 Mobile | 3.044 ms | 1 - 154 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-6690 | 53.718 ms | 1 - 275 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® SA7255P | 15.489 ms | 1 - 142 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Snapdragon® 8 Elite Mobile | 3.747 ms | 0 - 145 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® SA8295P | 10.967 ms | 1 - 142 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-7790 | 9.751 ms | 1 - 269 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ Q-8750 | 3.747 ms | 0 - 145 MB | NPU
| EfficientNet-B4 | QNN_DLC | w8a16 | Qualcomm® Dragonwing™ IQ-X7181 | 9.14 ms | 1 - 1 MB | NPU
| EfficientNet-B4 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 5.777 ms | 0 - 162 MB | NPU
| EfficientNet-B4 | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 21.925 ms | 0 - 206 MB | NPU
| EfficientNet-B4 | TFLITE | float | Qualcomm® QCS8275 | 28.916 ms | 0 - 97 MB | NPU
| EfficientNet-B4 | TFLITE | float | Qualcomm® Dragonwing™ QCS8550 (Proxy) | 8.019 ms | 0 - 3 MB | NPU
| EfficientNet-B4 | TFLITE | float | Qualcomm® SA8775P | 10.279 ms | 0 - 99 MB | NPU
| EfficientNet-B4 | TFLITE | float | Qualcomm® SA8650P | 10.279 ms | 0 - 99 MB | NPU
| EfficientNet-B4 | TFLITE | float | Qualcomm® SA8255P | 10.279 ms | 0 - 99 MB | NPU
| EfficientNet-B4 | TFLITE | float | Qualcomm® QCS8450 | 21.925 ms | 0 - 206 MB | NPU
| EfficientNet-B4 | TFLITE | float | Qualcomm® Dragonwing™ IQ-9075 | 9.996 ms | 0 - 49 MB | NPU
| EfficientNet-B4 | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 3.318 ms | 0 - 98 MB | NPU
| EfficientNet-B4 | TFLITE | float | Qualcomm® SA7255P | 28.916 ms | 0 - 97 MB | NPU
| EfficientNet-B4 | TFLITE | float | Snapdragon® 8 Elite Mobile | 4.313 ms | 0 - 105 MB | NPU
| EfficientNet-B4 | TFLITE | float | Qualcomm® SA8295P | 18.842 ms | 0 - 140 MB | NPU
| EfficientNet-B4 | TFLITE | float | Qualcomm® Dragonwing™ Q-8750 | 4.313 ms | 0 - 105 MB | NPU

## License
* The license for the original implementation of EfficientNet-B4 can be found
  [here](https://github.com/pytorch/vision/blob/main/LICENSE).

## References
* [EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks](https://arxiv.org/abs/1905.11946)
* [Source Model Implementation](https://github.com/pytorch/vision/blob/main/torchvision/models/efficientnet.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).