Open In Github

Python scripts for performing monocular depth estimation using the SC_Depth model in ONNX

Original image:https://commons.wikimedia.org/wiki/File:Cannery_District_Bozeman_Epic_Fitness_Interior_Wood_Stairs.jpg

Requirements

• Check the requirements.txt file.
• For ONNX, if you have a NVIDIA GPU, then install the onnxruntime-gpu, otherwise use the onnxruntime library.

Installation

git clone https://github.com/ibaiGorordo/ONNX-SCDepth-Monocular-Depth-Estimation.git
cd ONNX-YOLOv7-Object-Detection
pip install -r requirements.txt

ONNX Runtime

For Nvidia GPU computers: pip install onnxruntime-gpu

Otherwise: pip install onnxruntime

ONNX model

I don’t provide the model, but you can easily convert the Pytorch model by placing the following code in the line #80 of inference.py in the original repository:

model_name = "sc_depth_v3_nyu.onnx"
torch.onnx.export(model,  # model being run
                  tensor_img,  # model input (or a tuple for multiple inputs)
                  model_name,  # where to save the model (can be a file or file-like object)
                  export_params=True,  # store the trained parameter weights inside the model file
                  opset_version=16)

Then you can simplify the ONNX model using ONNX-Simplifier:

onnxsim sc_depth_v3_nyu.onnx sc_depth_v3_nyu_sim.onnx

Finally, copy the simplified ONNX model file to the models folder.

Original Pytorch model

The Pytorch pretrained models were taken from the original repository.

Examples

• Image inference:

 python image_depth_estimation.py 

• Video inference:

 python video_depth_estimation.py

• Webcam inference:

 python webcam_depth_estimation.py

Inference video Example

Original video: https://youtu.be/e0IjlkU-pX0

References:

• SC_Depth model: https://github.com/JiawangBian/sc_depth_pl
• PINTO0309’s model zoo: https://github.com/PINTO0309/PINTO_model_zoo
• PINTO0309’s model conversion tool: https://github.com/PINTO0309/openvino2tensorflow
• Original papers: https://arxiv.org/abs/2211.03660