VBGPS drone
==

## Build model (on server)

### Environment
:::info
Under `/Hierarchical-Localization/`

conda create --name [conda_server_name] python=3.7.11
conda activate [conda_server_name] `/Hloc/`
pip install -r requirements.txt

:::

### [Colmap](https://colmap.github.io/gui.html) installation

:::info

sudo apt-get install \
    git \
    cmake \
    ninja-build \
    build-essential \
    libboost-program-options-dev \
    libboost-filesystem-dev \
    libboost-graph-dev \
    libboost-system-dev \
    libboost-test-dev \
    libeigen3-dev \
    libflann-dev \
    libfreeimage-dev \
    libmetis-dev \
    libgoogle-glog-dev \
    libgflags-dev \
    libsqlite3-dev \
    libglew-dev \
    qtbase5-dev \
    libqt5opengl5-dev \
    libcgal-dev \
    libceres-dev

git clone https://github.com/colmap/colmap.git
cd colmap
git checkout dev
mkdir build
cd build
cmake .. -GNinja -DCMAKE_CUDA_ARCHITECTURES=sm_70
ninja
sudo ninja install

sudo apt-get install -y \
    nvidia-cuda-toolkit \
    nvidia-cuda-toolkit-gcc

:::

### Preparing datasets
:::info
Under `/Hierarchical-Localization/datasets/`
:::
- Create a folder in datasets
- Create three folder in your dataset folder: `db_images/` `images_upright/` `queries/`
- Put all images into `db_images/`
- Copy a intrinsic.txt (from `/1206_drone/queries/intrinsic.txt`) to `queries/`

$ mkdir [datasets_name]
$ cd [datasets_name]
$ mkdir db_images/ images_upright/ queries/
$ cp ../1206_drone/queries/intrinsic.txt queries/

- Edit intrinsic.txt (change the **width** and **height** to match your images in `./db_images`)

SIMPLE_RADIAL [images width] [images height] 512.015 480 360 -0.00169481

### SfM (Start to build)

:::info
Under `/Hierarchical-Localization/`
Environmemt: conda activate [conda_server_name]
:::

Construct 3 bins in `outputs/[datasets_name]/sfm_superpoint+NN`

python3 -m hloc.pipelines.VBgps.Sfm --name [datasets_name]

### Extract global feature

:::info
Location `/Hierarchical-Localization/`
Environmemt: conda activate [conda_server_name]
:::
Extract global feature from database

python3 -m hloc.pipelines.VBgps.extract_db_global --name [datasets_name]

Download the folder:
`/Hierarchical-Localization/outputs/[datasets_name]/sfm_superpoint+NN/`
`/Hierarchical-Localization/datasets/[datasets_name]/db_images/`

---

## Convert the model form

### Convert bin to nvm and Edit the Coordinate_demo.txt

:::info
Install [COLMAP](https://colmap.github.io/gui.html), [VisualSfM](http://ccwu.me/vsfm/index.html)
:::
Convert 3 bins to nvm by COLMAP
1. Open **COLMAP** GUI and import model
2. Choose the directory of `/sfm_superpoint+NN/`
3. Press open database
4. Select database.db in `/sfm_superpoint+NN/`
5. Select Images
6. Choose the directory of `/db_images/`
7. Find out your four fundamental images and *press* them
8. You can get their **qw qx qy qz tx ty tz**
9. Open `/VBGPS_coor_transform/Coordinate_demo.txt` and write it down (four images pose)
10. Press **Export model as**
11. Choose **.nvm** form and save it

### Convert nvm to ply

1. Open VisualSFM GUI and load the **.nvm** model
2. SfM -> Extra Functions -> Save Current Model
3. Save model as **[model_name].ply**
4. Open the **.ply** file and you may find many line include **(-1.#IND)**
5. Delete all the lines including **(-1.#IND)**
and move **.ply** file to `/VBGPS_client/Examples/Monocular/model/`

---

## Transform coordinate system

### Environment

:::info
Under `/VBGPS_coor_transform/`

conda create --name [conda_transform_name] python=3.9.12
conda activate [conda_transform_name]
pip install -r requirements.txt

:::

### Transform

You can edit the ***old_coor*** by yourself
(depand on the distance(m) between four fundenmental images pose)


Start to transform:

python3 Coordinate_Transform.py

You can get a **init.xml**, move it to `/VBGPS_client/Examples/Monocular/config/`

mv init.xml ../VBGPS_client/Examples/Monocular/config/

---

## Start to demo

:::info
Cuda == 8.0
OpenCV == 3.X
:::

### Convert ply to txt

:::info
Under `/VBGPS_client/Examples/Monocular/model/`
:::
Make sure the **.ply** file is under `/VBGPS_client/Examples/Monocular/model/`

./model.out [model_name].ply
mv model.txt [model_name].txt

### Compiling

:::info
Under `/VBGPS_client/`
:::

./build.sh

Must compiled successful without any error

### Build a config file (test.xml)

:::info
Under `/VBGPS_client/Examples/Monocular/config/`
:::
You can copy a `config_1206_drone_rm_IND.xml` as a `test.xml`

cp config_1206_drone_rm_IND.xml test.xml

And edit the `test.xml` change the `<MODEL_NAME>` to your model(.txt) path

### Run server (on server)

:::info
Under `/Hierarchical-Localization/`

conda activate [conda_server_name]

:::

Run server according to the sepecific datasets

python3 -m hloc.pipelines.VBgps.pipeline_vbgps --dataset datasets/[datasets_name] --outputs outputs/[datasets_name]

Make sure you get the scene

### Get Target image pose

:::info
Under `/VBGPS_client/Examples/Monocular/`
:::

./client_Ray_get_target ../../Vocabulary/ORBvoc.txt [Target_image_path] config/test.xml

Make sure you will get a `target.xml` under `/VBGPS_client/Examples/Monocular/config/`

### Run client

:::info
Under `/VBGPS_client/Examples/Monocular/`
:::
Start the client and specifiy a test video (Replace `test.xml` by you model config)

./client_Ray_cropped ../../Vocabulary/ORBvoc.txt video [test_video_path] config/test.xml config/init.xml config/target.xml

Start the client and demo on drone

./client_Ray_drone ../../Vocabulary/ORBvoc.txt config/test.xml config/init.xml config/target.xml