README.md

---
library_name: pytorch
license: other
tags:
- android
pipeline_tag: depth-estimation

---

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

# CREStereo: Optimized for Qualcomm Devices

CREStereo (Cascaded Recurrent Network with Adaptive Correlation) is a CVPR 2022 Oral paper that achieves state-of-the-art stereo matching accuracy.

This is based on the implementation of CREStereo found [here](https://github.com/ibaiGorordo/CREStereo-Pytorch).
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/main/src/qai_hub_models/models/crestereo) 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 |
|---|---|---|---|---|
| QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/crestereo/releases/v0.53.1/crestereo-qnn_dlc-float.zip)
| TFLITE | float | Universal |  | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/crestereo/releases/v0.53.1/crestereo-tflite-float.zip)

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


### Option 2: Export with Custom Configurations

Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/crestereo) 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 [CREStereo on GitHub](https://github.com/qualcomm/ai-hub-models/blob/main/src/qai_hub_models/models/crestereo) for usage instructions.

## Model Details

**Model Type:** Model_use_case.depth_estimation

**Model Stats:**
- Model checkpoint: CREStereo ETH3D pretrained (crestereo_eth3d.pt)
- Input: Rectified stereo pair — left and right RGB images
- Input resolution: 240x320
- Output: Disparity map
- Number of parameters: 5.43M
- Model size (float): 20.7 MB

## Performance Summary
| Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
|---|---|---|---|---|---|---
| CREStereo | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 3885.046 ms | 76 - 97 MB | CPU
| CREStereo | ONNX | float | Snapdragon® 8 Elite Mobile | 4065.807 ms | 63 - 86 MB | CPU
| CREStereo | ONNX | float | Snapdragon® X2 Elite | 1996.124 ms | 185 - 185 MB | CPU
| CREStereo | ONNX | float | Snapdragon® X Elite | 5407.067 ms | 180 - 180 MB | CPU
| CREStereo | ONNX | float | Snapdragon® X Elite | 5407.067 ms | 180 - 180 MB | CPU
| CREStereo | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 4685.376 ms | 62 - 82 MB | CPU
| CREStereo | ONNX | float | Qualcomm® QCS8550 (Proxy) | 5776.319 ms | 66 - 69 MB | CPU
| CREStereo | ONNX | float | Qualcomm® QCS9075 | 3982.804 ms | 94 - 97 MB | CPU
| CREStereo | ONNX | float | Snapdragon® 8 Elite For Galaxy Mobile | 4065.807 ms | 63 - 86 MB | CPU
| CREStereo | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 2886.412 ms | 144 - 235 MB | CPU
| CREStereo | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 3526.575 ms | 91 - 184 MB | CPU
| CREStereo | QNN_DLC | float | Snapdragon® X2 Elite | 2748.794 ms | 85 - 85 MB | CPU
| CREStereo | QNN_DLC | float | Snapdragon® X Elite | 8322.542 ms | 85 - 85 MB | CPU
| CREStereo | QNN_DLC | float | Snapdragon® X Elite | 8322.542 ms | 85 - 85 MB | CPU
| CREStereo | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 5099.593 ms | 127 - 221 MB | CPU
| CREStereo | QNN_DLC | float | Qualcomm® QCS8275 (Proxy) | 10856.074 ms | 130 - 216 MB | CPU
| CREStereo | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 5679.464 ms | 119 - 154 MB | CPU
| CREStereo | QNN_DLC | float | Qualcomm® SA8775P | 6657.907 ms | 128 - 214 MB | CPU
| CREStereo | QNN_DLC | float | Qualcomm® SA8775P | 6657.907 ms | 128 - 214 MB | CPU
| CREStereo | QNN_DLC | float | Qualcomm® SA8775P | 6657.907 ms | 128 - 214 MB | CPU
| CREStereo | QNN_DLC | float | Qualcomm® QCS9075 | 7730.53 ms | 298 - 1200 MB | CPU
| CREStereo | QNN_DLC | float | Qualcomm® QCS8450 (Proxy) | 7656.173 ms | 140 - 235 MB | CPU
| CREStereo | QNN_DLC | float | Qualcomm® SA7255P | 10856.074 ms | 130 - 216 MB | CPU
| CREStereo | QNN_DLC | float | Qualcomm® SA8295P | 5691.883 ms | 129 - 215 MB | CPU
| CREStereo | QNN_DLC | float | Snapdragon® 8 Elite For Galaxy Mobile | 3526.575 ms | 91 - 184 MB | CPU
| CREStereo | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 2707.927 ms | 46 - 73 MB | CPU
| CREStereo | TFLITE | float | Snapdragon® 8 Elite Mobile | 2871.556 ms | 47 - 74 MB | CPU
| CREStereo | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 4291.132 ms | 47 - 78 MB | CPU
| CREStereo | TFLITE | float | Qualcomm® QCS8275 (Proxy) | 7647.493 ms | 48 - 72 MB | CPU
| CREStereo | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 5070.01 ms | 47 - 59 MB | CPU
| CREStereo | TFLITE | float | Qualcomm® SA8775P | 5629.087 ms | 48 - 73 MB | CPU
| CREStereo | TFLITE | float | Qualcomm® SA8775P | 5629.087 ms | 48 - 73 MB | CPU
| CREStereo | TFLITE | float | Qualcomm® SA8775P | 5629.087 ms | 48 - 73 MB | CPU
| CREStereo | TFLITE | float | Qualcomm® QCS9075 | 5737.104 ms | 47 - 153 MB | CPU
| CREStereo | TFLITE | float | Qualcomm® QCS8450 (Proxy) | 6604.064 ms | 47 - 79 MB | CPU
| CREStereo | TFLITE | float | Qualcomm® SA7255P | 7647.493 ms | 48 - 72 MB | CPU
| CREStereo | TFLITE | float | Qualcomm® SA8295P | 3673.911 ms | 47 - 72 MB | CPU
| CREStereo | TFLITE | float | Snapdragon® 8 Elite For Galaxy Mobile | 2871.556 ms | 47 - 74 MB | CPU

## License
* The license for the original implementation of CREStereo can be found
  [here](https://github.com/megvii-research/CREStereo/blob/master/LICENSE).

## References
* [Practical Stereo Matching via Cascaded Recurrent Network with Adaptive Correlation](https://arxiv.org/abs/2203.11483)
* [Source Model Implementation](https://github.com/ibaiGorordo/CREStereo-Pytorch)

## 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).