|
--- |
|
license: mit |
|
library_name: unity-sentis |
|
pipeline_tag: depth-estimation |
|
tags: |
|
- sentis |
|
- unity |
|
- depth estimation |
|
--- |
|
|
|
These are [MiDaS](https://github.com/isl-org/MiDaS) models converted to ONNX to do Monocular Depth Estimation with Unity Sentis. |
|
|
|
## How to Use |
|
Import the package into Unity using [GitHub](https://github.com/julienkay/com.doji.midas), [OpenUPM](https://openupm.com/packages/com.doji.midas/) or the [Asset Store](https://assetstore.unity.com/packages/tools/ai-ml-integration/midas-monocular-depth-estimation-268501?aid=1101l3w5RJ&pubref=hf) |
|
|
|
Example source code to run this model can be found in the [Documentation](https://docs.doji-tech.com/com.doji.midas/manual/getting-started.html) |
|
|
|
 |
|
|
|
## Model Details |
|
The MiDaS models below were converted to ONNX using [this colab notebook](https://github.com/julienkay/com.doji.midas/blob/v1.0.0/tools/MiDaS_ONNX_Export.ipynb) |
|
Input normalization is baked into the models, so the model input 'input_image' is expected to be in the [0,1] range. |
|
Input sizes are static (e.g. 256, 384 or 512 denoted by the model suffix) |
|
|
|
## Unity Sentis |
|
Unity Sentis is the inference engine that runs in Unity 3D. More information can be found [here](https://unity.com/products/sentis) |
|
|
