Newer
Older
This project is used to compare state-of-the-art stereo RGB-D cameras from two manufactures, namely Realsense and ZED.
Together with evaluation of several models; MediaPipe and ViTPose.
The goal is to find the best camera for our use case.
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
Before we start, we need to install ZED SDK and Realsense SDK as follows:
### ZED SDK
1. Download ZED SDK from [here](https://www.stereolabs.com/developers/release/).
* Be aware, that ZED SDK is supported only together with CUDA. Therefore, only with NVIDIA GPU.
* If you don't have NVIDIA GPU, you can still jump to Realsense SDK installation.
2. Install ZED SDK according to the [documentation](https://www.stereolabs.com/docs/app-development/python/install/).
* It is important that you build the wheels for your python version and architecture.
* After the installation, you should see in the current directory three built wheels similarly like this:
```
PyOpenGL_accelerate-[VERSION]-cp[PYTHON_VERSION]-cp[PYTHON_VERSION]-win_amd64.whl
PyOpenGL-[VERSION]-cp[PYTHON_VERSION]-cp[PYTHON_VERSION]-win_amd64.whl
pyzed-[VERSION]-cp[PYTHON_VERSION]-cp[PYTHON_VERSION]-win_amd64.whl
```
* During the SDK installation, you also built binaries of such SDK. Such binaries are necessary for later usage in your other virtual environment.
* To work with new virtual environment, use the following steps:
* Create your new virtual environment, e.g. `python -m venv .venv`
* Activate your new virtual environment, e.g. `.venv/Scripts/activate`
* Install **ALL** the built wheels, e.g. `pip install PyOpenGL_accelerate-[VERSION]-cp[PYTHON_VERSION]-cp[PYTHON_VERSION]-win_amd64.whl`
* Find the binaries in your ZED SDK installation folder.
* They are usually located in `ZED_SDK/bin/`
* Based on your HW, they can be found as e.g. `ZED_SDK/bin/sl_ai64.dll` and `ZED_SDK/bin/sl_zed64.dll`
* Copy them into your new virtual environment, e.g. `.venv/Lib/site-packages/pyzed/`
* After that, you should be ready to go...
### Realsense SDK
1. Follow the [documentation](https://www.intelrealsense.com/sdk-2/) to install Realsense SDK.
2. The easiest way to download SDK into your virtual environment is through [pip](https://www.w3schools.com/python/python_pip.asp). `pip install pyrealsense2`
## Setup the environment
For better reproducibility, we also provide `requierments.txt` file with all necessary dependencies.
1. After you clone our repository
2. Install dependencies in root directory with `pip install -r requirements.txt`
3. If any package fails, look for version mismatch between your python version and its package version.
## Run
In order to run the application, you can choose from several scripts from pure streaming `streaming/cameras/camera.py`
to running mediapipe or zed native `detection/pipelines/` or evaluation scripts upon CSV files in `evaluation/`.
Because the hierarchy of the project is quite complex and was used primarily for Thesis purposes, we recommend to use
contact the author for further guidance.