### Install ByteTrack on Host Machine

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Clone the repository, install dependencies, and set up ByteTrack for development.

```shell
git clone https://github.com/ifzhang/ByteTrack.git
cd ByteTrack
pip3 install -r requirements.txt
python3 setup.py develop
```

--------------------------------

### Run BYTE Tracker Example

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/cstrack/README.md

Execute the BYTE tracker example script with specified configuration and model paths. Ensure you replace 'your/data/path' with the actual path to your data.

```bash
cd tracking
python3 test_cstrack.py --val_mot17 True --val_hf 2 --weights ../weights/mix_mot17_half_cstrack.pt --conf_thres 0.6 --data_cfg ../src/lib/cfg/mot17_hf.json --data_dir your/data/path
```

--------------------------------

### Install cython_bbox

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Install the cython_bbox package.

```shell
pip3 install cython_bbox
```

--------------------------------

### Install Eigen Library

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/TensorRT/cpp/README.md

Steps to download, unzip, and install the Eigen library for C++ projects.

```shell
unzip eigen-3.3.9.zip
cd eigen-3.3.9
mkdir build
cd build
cmake ..
sudo make install
```

--------------------------------

### Include and Link Directories Setup

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/TensorRT/cpp/CMakeLists.txt

Specifies include and link directories for project sources, Eigen, CUDA, cuDNN, and TensorRT. Adapt paths if your installations differ.

```cmake
include_directories(${PROJECT_SOURCE_DIR}/include)
include_directories(/usr/local/include/eigen3)
link_directories(${PROJECT_SOURCE_DIR}/include)
# include and link dirs of cuda and tensorrt, you need adapt them if yours are different
# cuda
include_directories(/usr/local/cuda/include)
link_directories(/usr/local/cuda/lib64)
# cudnn
include_directories(/data/cuda/cuda-10.2/cudnn/v8.0.4/include)
link_directories(/data/cuda/cuda-10.2/cudnn/v8.0.4/lib64)
# tensorrt
include_directories(/opt/tiger/demo/TensorRT-7.2.3.4/include)
link_directories(/opt/tiger/demo/TensorRT-7.2.3.4/lib)
```

--------------------------------

### CMake Project Setup

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/DeepStream/CMakeLists.txt

Configures the CMake module path, minimum version, CUDA compiler, C++ standard, and library output directory.

```cmake
set(CMAKE_MODULE_PATH ${CMAKE_CURRENT_SOURCE_DIR}/cmake)
cmake_minimum_required(VERSION 3.16)
set(CMAKE_CUDA_COMPILER /usr/local/cuda/bin/nvcc)
set(CMAKE_CXX_STANDARD 14)
set(CMAKE_LIBRARY_OUTPUT_DIRECTORY ${CMAKE_SOURCE_DIR}/lib)
```

--------------------------------

### Configure OpenCV and Build Examples

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/ncnn/cpp/CMakeLists.txt

Configures OpenCV finding and builds ncnn examples if OpenCV is found or NCNN_SIMPLEOCV is enabled. It sets C++ standard and includes necessary directories.

```cmake
if(NCNN_PIXEL)
    find_package(OpenCV QUIET COMPONENTS opencv_world)
    # for opencv 2.4 on ubuntu 16.04, there is no opencv_world but OpenCV_FOUND will be TRUE
    if("${OpenCV_LIBS}" STREQUAL "")
        set(OpenCV_FOUND FALSE)
    endif()
    if(NOT OpenCV_FOUND)
        find_package(OpenCV QUIET COMPONENTS core highgui imgproc imgcodecs videoio)
    endif()
    if(NOT OpenCV_FOUND)
        find_package(OpenCV QUIET COMPONENTS core highgui imgproc)
    endif()

    if(OpenCV_FOUND OR NCNN_SIMPLEOCV)
        if(OpenCV_FOUND)
            message(STATUS "OpenCV library: ${OpenCV_INSTALL_PATH}")
            message(STATUS "    version: ${OpenCV_VERSION}")
            message(STATUS "    libraries: ${OpenCV_LIBS}")
            message(STATUS "    include path: ${OpenCV_INCLUDE_DIRS}")

            if(${OpenCV_VERSION_MAJOR} GREATER 3)
                set(CMAKE_CXX_STANDARD 11)
            endif()
        endif()

        include_directories(${CMAKE_CURRENT_SOURCE_DIR}/../src)
        include_directories(${CMAKE_CURRENT_BINARY_DIR}/../src)
        include_directories(include)
        include_directories(/usr/local/include/eigen3)

        ncnn_add_example(squeezenet)
        ncnn_add_example(squeezenet_c_api)
        ncnn_add_example(fasterrcnn)
        ncnn_add_example(rfcn)
        ncnn_add_example(yolov2)
        ncnn_add_example(yolov3)
        if(OpenCV_FOUND)
            ncnn_add_example(yolov4)
        endif()
        ncnn_add_example(yolov5)
        ncnn_add_example(yolox)
        ncnn_add_example(mobilenetv2ssdlite)
        ncnn_add_example(mobilenetssd)
        ncnn_add_example(squeezenetssd)
        ncnn_add_example(shufflenetv2)
        ncnn_add_example(peleenetssd_seg)
        ncnn_add_example(simplepose)
        ncnn_add_example(retinaface)
        ncnn_add_example(yolact)
        ncnn_add_example(nanodet)
        ncnn_add_example(scrfd)
        ncnn_add_example(scrfd_crowdhuman)
        ncnn_add_example(rvm)
        file(GLOB My_Source_Files src/*.cpp)
        add_executable(bytetrack ${My_Source_Files})
        if(OpenCV_FOUND)
            target_include_directories(bytetrack PRIVATE ${OpenCV_INCLUDE_DIRS})
            target_link_libraries(bytetrack PRIVATE ncnn ${OpenCV_LIBS})
        elseif(NCNN_SIMPLEOCV)
            target_compile_definitions(bytetrack PUBLIC USE_NCNN_SIMPLEOCV)
            target_link_libraries(bytetrack PRIVATE ncnn)
        endif()
        # add test to a virtual project group
        set_property(TARGET bytetrack PROPERTY FOLDER "examples")
    else()
        message(WARNING "OpenCV not found and NCNN_SIMPLEOCV disabled, examples won't be built")
    endif()
else()
    message(WARNING "NCNN_PIXEL not enabled, examples won't be built")
endif()
```

--------------------------------

### Install pycocotools

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Install the pycocotools library, which is a dependency for some datasets.

```shell
pip3 install cython; pip3 install 'git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI'
```

--------------------------------

### Run ByteTrack Docker Container

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Start a Docker container with GPU support, mounting local directories for pretrained models, datasets, and output logs. This command also configures display and device access for interactive use.

```shell
mkdir -p pretrained && \
mkdir -p YOLOX_outputs && \
xhost +local: && \
docker run --gpus all -it --rm \
-v $PWD/pretrained:/workspace/ByteTrack/pretrained \
-v $PWD/datasets:/workspace/ByteTrack/datasets \
-v $PWD/YOLOX_outputs:/workspace/ByteTrack/YOLOX_outputs \
-v /tmp/.X11-unix/:/tmp/.X11-unix:rw \
--device /dev/video0:/dev/video0:mwr \
--net=host \
-e XDG_RUNTIME_DIR=$XDG_RUNTIME_DIR \
-e DISPLAY=$DISPLAY \
--privileged \
bytetrack:latest
```

--------------------------------

### Define ncnn_add_example Macro

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/ncnn/cpp/CMakeLists.txt

Defines a CMake macro to simplify adding ncnn example executables. It handles executable creation, OpenCV linking, and setting target properties.

```cmake
macro(ncnn_add_example name)
    add_executable(${name} ${name}.cpp)
    if(OpenCV_FOUND)
        target_include_directories(${name} PRIVATE ${OpenCV_INCLUDE_DIRS})
        target_link_libraries(${name} PRIVATE ncnn ${OpenCV_LIBS})
    elseif(NCNN_SIMPLEOCV)
        target_compile_definitions(${name} PUBLIC USE_NCNN_SIMPLEOCV)
        target_link_libraries(${name} PRIVATE ncnn)
    endif()

    # add test to a virtual project group
    set_property(TARGET ${name} PROPERTY FOLDER "examples")
endmacro()
```

--------------------------------

### Optimize Running Speed

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/ncnn/cpp/README.md

Example of modifying the 'num_threads' variable in bytetrack.cpp to optimize performance based on CPU cores.

```cpp
yolox.opt.num_threads = 20;

```

--------------------------------

### Train Custom Dataset

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Train ByteTrack on a custom dataset. Prepare your data in COCO format, create a custom Exp file referencing existing examples, and modify get_data_loader and get_eval_loader functions.

```shell
cd <ByteTrack_HOME>
python3 tools/train.py -f exps/example/mot/your_exp_file.py -d 8 -b 48 --fp16 -o -c pretrained/yolox_x.pth
```

--------------------------------

### Modify ncnn Param File

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/ncnn/cpp/README.md

Example of modifying the bytetrack_s.param file to incorporate the YoloV5Focus layer, necessary for ncnn compatibility.

```text
235 268
Input            images                   0 1 images
Split            splitncnn_input0         1 4 images images_splitncnn_0 images_splitncnn_1 images_splitncnn_2 images_splitncnn_3
Crop             Slice_4                  1 1 images_splitncnn_3 467 -23309=1,0 -23310=1,2147483647 -23311=1,1
Crop             Slice_9                  1 1 467 472 -23309=1,0 -23310=1,2147483647 -23311=1,2
Crop             Slice_14                 1 1 images_splitncnn_2 477 -23309=1,0 -23310=1,2147483647 -23311=1,1
Crop             Slice_19                 1 1 477 482 -23309=1,1 -23310=1,2147483647 -23311=1,2
Crop             Slice_24                 1 1 images_splitncnn_1 487 -23309=1,1 -23310=1,2147483647 -23311=1,1
Crop             Slice_29                 1 1 487 492 -23309=1,0 -23310=1,2147483647 -23311=1,2
Crop             Slice_34                 1 1 images_splitncnn_0 497 -23309=1,1 -23310=1,2147483647 -23311=1,1
Crop             Slice_39                 1 1 497 502 -23309=1,1 -23310=1,2147483647 -23311=1,2
Concat           Concat_40                4 1 472 492 482 502 503 0=0
...

```

```text
226 328
Input            images                   0 1 images
YoloV5Focus      focus                    1 1 images 503
...

```

--------------------------------

### Run ByteTrack Demo

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/ncnn/cpp/README.md

Command to execute the ByteTrack C++ demo with a video file.

```shell
./bytetrack palace.mp4
```

--------------------------------

### Build ByteTrack C++ Project

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/ncnn/cpp/README.md

Steps to copy necessary files into the ncnn/examples directory and build the ByteTrack C++ executable.

```shell
cd ncnn/build/examples
cmake ..
make
```

--------------------------------

### Run ByteTrack Demo

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Execute the ByteTrack demo script to perform tracking on a video. This command requires specifying the experiment configuration, checkpoint, and optional arguments like FP16 and fuse.

```shell
cd <ByteTrack_HOME>
python3 tools/demo_track.py video -f exps/example/mot/yolox_x_mix_det.py -c pretrained/bytetrack_x_mot17.pth.tar --fp16 --fuse --save_result
```

--------------------------------

### Run ByteTrack-TensorRT C++ Demo

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/TensorRT/cpp/README.md

Execute the compiled ByteTrack-TensorRT C++ demo with a specified engine file and input video.

```shell
./bytetrack ../../../../YOLOX_outputs/yolox_s_mix_det/model_trt.engine -i ../../../../videos/palace.mp4
```

--------------------------------

### Run QDTrack with ReID Motion

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/qdtrack/README.md

Execute the QDTrack ReID motion tracker using a specified configuration and pre-trained model. Ensure the distributed training environment is set up correctly.

```bash
python3 -m torch.distributed.launch --nproc_per_node=8 --master_port=29501 tools/test.py configs/mot17/qdtrack-frcnn_r50_fpn_4e_mot17.py work_dirs/mot17_half_qdtrack.pth --launcher pytorch --eval track --eval-options resfile_path=output
```

--------------------------------

### Build ByteTrack-TensorRT C++ Demo

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/TensorRT/cpp/README.md

Commands to build the ByteTrack-TensorRT C++ demo after setting up paths in CMakeLists.txt.

```shell
cd <ByteTrack_HOME>/deploy/TensorRT/cpp
mkdir build
cd build
cmake ..
make
```

--------------------------------

### Run ONNX Runtime Demo

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/ONNXRuntime/README.md

Executes the ONNX Runtime inference demo. This command should be run from the ONNXRuntime deployment directory within ByteTrack. The demo is capable of achieving approximately 16 FPS on a specified CPU configuration.

```shell
cd <ByteTrack_HOME>/deploy/ONNXRuntime
python3 onnx_inference.py
```

--------------------------------

### Build ByteTrack Docker Image

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Build a Docker image tagged as 'bytetrack:latest' from the current directory.

```shell
docker build -t bytetrack:latest .
```

--------------------------------

### Run TensorRT ByteTrack Demo

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/TensorRT/python/README.md

Executes a TensorRT-accelerated demo for ByteTrack object tracking using a pre-converted TensorRT model. This command enables TensorRT acceleration for higher FPS.

```shell
cd <ByteTrack_HOME>
python3 tools/demo_track.py video -f exps/example/mot/yolox_s_mix_det.py --trt --save_result
```

--------------------------------

### Convert Input Video for TensorRT C++ Demo

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/TensorRT/cpp/README.md

Utility command to convert input videos to a format compatible with the TensorRT C++ demo if frame loss occurs.

```shell
cd <ByteTrack_HOME>
python3 tools/convert_video.py
```

--------------------------------

### Configure Input Dimensions

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/TensorRT/cpp/README.md

Set the input width and height for the ByteTrack model in C++ source code.

```c++
static const int INPUT_W = 1088;
static const int INPUT_H = 608;
```

--------------------------------

### Run ByteTrack on MOT20 Test Set

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Execute ByteTrack for testing on the MOT20 dataset with specific parameters for matching threshold and MOT20 mode. The input size should be adjusted in the configuration file as noted.

```shell
cd <ByteTrack_HOME>
python3 tools/track.py -f exps/example/mot/yolox_x_mix_mot20_ch.py -c pretrained/bytetrack_x_mot20.pth.tar -b 1 -d 1 --fp16 --fuse --match_thresh 0.7 --mot20
python3 tools/interpolation.py
```

--------------------------------

### Finding Packages

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/DeepStream/CMakeLists.txt

Locates necessary packages like GLib and OpenCV for the project.

```cmake
find_package(GLib)
find_package(OpenCV)
```

--------------------------------

### Mix Datasets for Training and Testing

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Prepare data for training and testing by mixing different datasets. This involves running specific mix scripts after datasets have been converted to COCO format.

```shell
cd <ByteTrack_HOME>
python3 tools/mix_data_ablation.py
python3 tools/mix_data_test_mot17.py
python3 tools/mix_data_test_mot20.py
```

--------------------------------

### Include Directories

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/DeepStream/CMakeLists.txt

Adds include paths for GLib, OpenCV, DeepStream, and project-specific headers.

```cmake
include_directories(${GLIB2_INCLUDE_DIR})
include_directories(${OpenCV_INCLUDE_DIRS})
include_directories(${DS_INCLUDES_DIR})
include_directories(${CMAKE_SOURCE_DIR}/includes)
```

--------------------------------

### Run BYTE Tracker with FairMOT

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/fairmot/README.md

Execute the BYTE tracker script for MOT tasks. This configuration uses a match threshold of 0.8 and specifies the model path.

```python
python3 track_half.py mot --load_model ../exp/mot/mot17_half_dla34/model_last.pth --match_thres 0.8
```

--------------------------------

### Run JDE Tracker

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/jde/README.md

Execute the JDE tracking script with specified configuration and model weights. Ensure the model file is placed in the correct directory.

```bash
python3 track_half.py --cfg ./cfg/yolov3_1088x608.cfg --weights weights/mix_mot17_half_jde.pt
```

--------------------------------

### Basic CMake Configuration

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/TensorRT/cpp/CMakeLists.txt

Sets the minimum CMake version, project name, and C++ standard. It also configures build type and enables CUDA usage.

```cmake
cmake_minimum_required(VERSION 2.6)

project(bytetrack)

add_definitions(-std=c++11)

option(CUDA_USE_STATIC_CUDA_RUNTIME OFF)
set(CMAKE_CXX_STANDARD 11)
set(CMAKE_BUILD_TYPE Debug)

find_package(CUDA REQUIRED)
```

--------------------------------

### Run TransTrack

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/transtrack/README.md

Execute the TransTrack script with specified parameters for tracking.

```bash
python3 main_track.py  --output_dir . --dataset_file mot --coco_path mot --batch_size 1 --resume pretrained/671mot17_crowdhuman_mot17.pth --eval --with_box_refine --num_queries 500
```

--------------------------------

### DeepStream Configuration for ByteTracker

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/DeepStream/README.md

Configuration snippet for deepstream_app_config.txt to enable and specify the ByteTracker library. Ensure the ll-lib-file path is correct.

```ini
[tracker]
enable=1
tracker-width=640
tracker-height=384
gpu-id=0
ll-lib-file=//opt/nvidia/deepstream/deepstream/lib/libByteTracker.so
enable-batch-process=1
```

--------------------------------

### Executable Definition and Linking

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/TensorRT/cpp/CMakeLists.txt

Finds all C++ source files in the 'src' directory and creates the 'bytetrack' executable. Links necessary libraries including TensorRT, CUDA runtime, and OpenCV.

```cmake
file(GLOB My_Source_Files ${PROJECT_SOURCE_DIR}/src/*.cpp)
add_executable(bytetrack ${My_Source_Files})
target_link_libraries(bytetrack nvinfer)
target_link_libraries(bytetrack cudart)
target_link_libraries(bytetrack ${OpenCV_LIBS})
```

--------------------------------

### Build ByteTracker Library

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/DeepStream/README.md

Commands to build the ByteTracker C++ library. This creates a shared object file that can be used with DeepStream.

```bash
$mkdir build && cd build  

$cmake ..  

$make ByteTracker  
```

--------------------------------

### Train CTracker Model

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/ctracker/README.md

Run this command to train the CTracker model. Ensure the dataset is prepared and the generate_half_csv.py script has been executed.

```bash
python3 train.py --root_path MOT17 --csv_train train_half_annots.csv --model_dir ctracker/ --depth 50 --epochs 50
```

--------------------------------

### Train MOTR Model

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/motr/README.md

Command to initiate the training process for the MOTR model using distributed PyTorch. Ensure all specified paths and pretrained models are correctly set up.

```bash
python3 -m torch.distributed.launch --nproc_per_node=8 \
    --use_env main.py \
    --meta_arch motr \
    --dataset_file e2e_joint \
    --epoch 50 \
    --with_box_refine \
    --lr_drop 40 \
    --lr 2e-4 \
    --lr_backbone 2e-5 \
    --pretrained coco_model_final.pth \
    --output_dir exps/e2e_motr_r50_mot17trainhalf \
    --batch_size 1 \
    --sample_mode 'random_interval' \
    --sample_interval 10 \
    --sampler_steps 10 20 30 \
    --sampler_lengths 2 3 4 5 \
    --update_query_pos \
    --merger_dropout 0 \
    --dropout 0 \
    --random_drop 0.1 \
    --fp_ratio 0.3 \
    --query_interaction_layer 'QIM' \
    --extra_track_attn \
    --mot_path .
    --data_txt_path_train ./datasets/data_path/mot17.half \
    --data_txt_path_val ./datasets/data_path/mot17.val \

```

--------------------------------

### Test CTracker_BYTE Model

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/ctracker/README.md

Run this command to test the CTracker_BYTE model. This requires adding mot_online and byte_tracker.py to the CTracker repository, and using the test_byte.py script.

```bash
python3 test_byte.py --dataset_path MOT17 --model_dir ctracker --model_path ctracker/mot17_half_ctracker.pt
```

--------------------------------

### Run ByteTrack on MOT17 Validation Set

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Execute the ByteTrack algorithm for evaluation on the MOT17 validation set. Ensure you are in the ByteTrack home directory and have the necessary configuration and model files.

```shell
cd <ByteTrack_HOME>
python3 tools/track.py -f exps/example/mot/yolox_x_ablation.py -c pretrained/bytetrack_ablation.pth.tar -b 1 -d 1 --fp16 --fuse
```

--------------------------------

### Run ByteTrack on MOT17 Test Set

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Execute ByteTrack for testing on the MOT17 dataset. This command also includes running an interpolation script, likely for post-processing results before submission.

```shell
cd <ByteTrack_HOME>
python3 tools/track.py -f exps/example/mot/yolox_x_mix_det.py -c pretrained/bytetrack_x_mot17.pth.tar -b 1 -d 1 --fp16 --fuse
python3 tools/interpolation.py
```

--------------------------------

### Run TraDeS_BYTE Tracking

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/trades/README.md

Execute the TraDeS_BYTE tracking script with specified parameters. This version uses byte_tracker.py for motion tracking with a Kalman filter.

```bash
python3 test.py tracking --exp_id mot17_half --dataset mot --dataset_version 17halfval --pre_hm --ltrb_amodal --inference --load_model ../models/mot_half.pth --gpus 0 --clip_len 3 --trades  --track_thresh 0.4 --new_thresh 0.5 --out_thresh 0.1 --pre_thresh 0.5
```

--------------------------------

### Additional Build Definitions

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/TensorRT/cpp/CMakeLists.txt

Applies additional build definitions for optimization and threading.

```cmake
add_definitions(-O2 -pthread)
```

--------------------------------

### Run TraDeS Tracking

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/trades/README.md

Execute the TraDeS tracking script with specified parameters. Ensure the model and dataset paths are correct.

```bash
python3 test.py tracking --exp_id mot17_half --dataset mot --dataset_version 17halfval --pre_hm --ltrb_amodal --inference --load_model ../models/mot_half.pth --gpus 0 --clip_len 3 --trades --track_thresh 0.4 --new_thresh 0.4 --out_thresh 0.2 --pre_thresh 0.5
```

--------------------------------

### Test CTracker Model

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/ctracker/README.md

Execute this command to test the trained CTracker model. Make sure to replace the test.py script with the one from the CTracker repository.

```bash
python3 test.py --dataset_path MOT17 --model_dir ctracker --model_path ctracker/mot17_half_ctracker.pt
```

--------------------------------

### Generate ncnn Param and Bin Files

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/ncnn/cpp/README.md

Convert the generated ONNX file into ncnn's param and bin formats using the onnx2ncnn tool.

```shell
cd ncnn/build/tools/onnx
./onnx2ncnn bytetrack_s.onnx bytetrack_s.param bytetrack_s.bin
```

--------------------------------

### Run Other Trackers on MOT17 Validation Set

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Compare ByteTrack performance by running SORT, DeepSORT, and MOTDT trackers. These commands use the same configuration and model as the ByteTrack run.

```shell
python3 tools/track_sort.py -f exps/example/mot/yolox_x_ablation.py -c pretrained/bytetrack_ablation.pth.tar -b 1 -d 1 --fp16 --fuse
```

```shell
python3 tools/track_deepsort.py -f exps/example/mot/yolox_x_ablation.py -c pretrained/bytetrack_ablation.pth.tar -b 1 -d 1 --fp16 --fuse
```

```shell
python3 tools/track_motdt.py -f exps/example/mot/yolox_x_ablation.py -c pretrained/bytetrack_ablation.pth.tar -b 1 -d 1 --fp16 --fuse
```

--------------------------------

### Convert PyTorch Model to ONNX

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/ONNXRuntime/README.md

Converts a PyTorch model to ONNX format using the provided export script. Ensure you are in the ByteTrack home directory and specify the output name, configuration file, and pre-trained weights.

```shell
cd <ByteTrack_HOME>
python3 tools/export_onnx.py --output-name bytetrack_s.onnx -f exps/example/mot/yolox_s_mix_det.py -c pretrained/bytetrack_s_mot17.pth.tar
```

--------------------------------

### Optimize ncnn Model

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/ncnn/cpp/README.md

Use ncnnoptimize to optimize the param and bin files for better performance.

```shell
../ncnnoptimize bytetrack_s.param bytetrack_s.bin bytetrack_s_op.param bytetrack_s_op.bin 65536
```

--------------------------------

### Train MOT20 Test Model

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Train the MOT20 test model using MOT20 train and CrowdHuman datasets. This command requires clipping bounding boxes within the image, with modifications suggested in provided file lines.

```shell
cd <ByteTrack_HOME>
python3 tools/train.py -f exps/example/mot/yolox_x_mix_mot20_ch.py -d 8 -b 48 --fp16 -o -c pretrained/yolox_x.pth
```

--------------------------------

### Replace CSTrack File

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/cstrack/README.md

Replace the existing cstrack.py file with the one from the specified GitHub URL.

```bash
replace https://github.com/JudasDie/SOTS/blob/master/lib/tracker/cstrack.py
```

--------------------------------

### Integrate BYTETracker with Other Detectors

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Pass detection results from other detectors to BYTETracker for tracking. Ensure detection results are formatted correctly for the tracker.

```python
from yolox.tracker.byte_tracker import BYTETracker
tracker = BYTETracker(args)
for image in images:
   dets = detector(image)
   online_targets = tracker.update(dets, info_imgs, img_size)
```

--------------------------------

### Run CenterTrack Tracking Test

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/centertrack/README.md

Execute the tracking test script for CenterTrack with specified experiment parameters and model.

```bash
python3 test.py tracking --exp_id mot17_half --dataset mot --dataset_version 17halfval --pre_hm --ltrb_amodal  --load_model ../models/mot17_half.pth --track_thresh 0.4 --new_thresh 0.5 --out_thresh 0.2 --pre_thresh 0.5
```

--------------------------------

### Generate ONNX File

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/ncnn/cpp/README.md

Command to generate the ONNX file from a PyTorch model checkpoint using the provided export script.

```shell
cd <ByteTrack_HOME>
python3 tools/export_onnx.py -f exps/example/mot/yolox_s_mix_det.py -c pretrained/bytetrack_s_mot17.pth.tar
```

--------------------------------

### Convert ETHZ to COCO Format

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Convert the ETHZ dataset to COCO format using the provided script. Ensure you are in the ByteTrack home directory.

```shell
cd <ByteTrack_HOME>
python3 tools/convert_ethz_to_coco.py
```

--------------------------------

### Source File Globbing

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/DeepStream/CMakeLists.txt

Collects all .cpp files from the src directory for compilation.

```cmake
file(GLOB TRACK ${CMAKE_SOURCE_DIR}/src/*.cpp)
```

--------------------------------

### Clone SOTS Repository

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/cstrack/README.md

Clone the SOTS repository from GitHub, specifically checking out the MOT branch.

```bash
git clone -b MOT https://github.com/JudasDie/SOTS.git
```

--------------------------------

### Convert PyTorch Model to TensorRT

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/TensorRT/python/README.md

Converts a PyTorch ByteTrack model to a TensorRT-compatible format. Ensure you are in the ByteTrack home directory and have the necessary configuration and model files.

```shell
cd <ByteTrack_HOME>
python3 tools/trt.py -f exps/example/mot/yolox_s_mix_det.py -c pretrained/bytetrack_s_mot17.pth.tar
```

--------------------------------

### Clone CenterTrack Repository

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/centertrack/README.md

Use git to clone the CenterTrack repository from GitHub.

```bash
git clone https://github.com/xingyizhou/CenterTrack.git
```

--------------------------------

### Compiler Flags and OpenCV Integration

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/TensorRT/cpp/CMakeLists.txt

Sets C++ compiler flags for optimization and error handling, and finds the OpenCV package, including its include directories.

```cmake
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11 -Wall -Ofast -Wfatal-errors -D_MWAITXINTRIN_H_INCLUDED")

find_package(OpenCV)
include_directories(${OpenCV_INCLUDE_DIRS})
```

--------------------------------

### Convert MOT20 to COCO Format

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Convert the MOT20 dataset to COCO format using the provided script. Ensure you are in the ByteTrack home directory.

```shell
cd <ByteTrack_HOME>
python3 tools/convert_mot20_to_coco.py
```

--------------------------------

### Library Linking

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/DeepStream/CMakeLists.txt

Specifies the directory for linking libraries and adds the ByteTracker shared library, linking it with nvds_meta.

```cmake
link_directories(${DS_LIBS_DIR})
add_library(ByteTracker SHARED ${TRACK})
target_link_libraries(ByteTracker nvds_meta)
```

--------------------------------

### Train Ablation Model (MOT17 half train and CrowdHuman)

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Use this command to train an ablation model using MOT17 half training data and CrowdHuman dataset. Ensure you have downloaded the COCO pretrained YOLOX model.

```shell
cd <ByteTrack_HOME>
python3 tools/train.py -f exps/example/mot/yolox_x_ablation.py -d 8 -b 48 --fp16 -o -c pretrained/yolox_x.pth
```

--------------------------------

### Evaluate CTracker Model

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/ctracker/README.md

Use this command to evaluate the CTracker model's performance on the MOT challenge. Ensure eval_motchallenge.py is added to the repository and ground truth files are prepared.

```bash
python3 eval_motchallenge.py --groundtruths MOT17/train --tests ctracker/results --gt_type half_val --eval_official  --score_threshold -1
```

--------------------------------

### DeepStream and Library Paths

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/DeepStream/CMakeLists.txt

Defines the root directory for DeepStream and specifies paths for its libraries and includes.

```cmake
set(DS_ROOT_DIR /opt/nvidia/deepstream/deepstream)
set(DS_LIBS_DIR ${DS_ROOT_DIR}/lib)
set(DS_INCLUDES_DIR ${DS_ROOT_DIR}/sources/includes)
```

--------------------------------

### Convert Cityperson to COCO Format

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Convert the Cityperson dataset to COCO format using the provided script. Ensure you are in the ByteTrack home directory.

```shell
cd <ByteTrack_HOME>
python3 tools/convert_cityperson_to_coco.py
```

--------------------------------

### Train MOT17 Test Model

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Command for training the MOT17 test model, incorporating MOT17 training data, CrowdHuman, Cityperson, and ETHZ datasets. Requires COCO pretrained YOLOX model.

```shell
cd <ByteTrack_HOME>
python3 tools/train.py -f exps/example/mot/yolox_x_mix_det.py -d 8 -b 48 --fp16 -o -c pretrained/yolox_x.pth
```

--------------------------------

### Convert MOT17 to COCO Format

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Convert the MOT17 dataset to COCO format using the provided script. Ensure you are in the ByteTrack home directory.

```shell
cd <ByteTrack_HOME>
python3 tools/convert_mot17_to_coco.py
```

--------------------------------

### Evaluate MOTR Model

Source: https://github.com/foundationvision/bytetrack/blob/main/tutorials/motr/README.md

Command to run the evaluation script for the MOTR model. This command is used after training to assess the model's performance on a validation set. Ensure the pretrained model path is correct.

```bash
python3 eval.py \
    --meta_arch motr \
    --dataset_file e2e_joint \
    --epoch 200 \
    --with_box_refine \
    --lr_drop 100 \
    --lr 2e-4 \
    --lr_backbone 2e-5 \
    --pretrained exps/e2e_motr_r50_mot17val/motr_final.pth \
    --output_dir exps/e2e_motr_r50_mot17val \
    --batch_size 1 \
    --sample_mode 'random_interval' \
    --sample_interval 10 \
    --sampler_steps 50 90 120 \
    --sampler_lengths 2 3 4 5 \
    --update_query_pos \
    --merger_dropout 0 \
    --dropout 0 \
    --random_drop 0.1 \
    --fp_ratio 0.3 \
    --query_interaction_layer 'QIM' \
    --extra_track_attn \
    --mot_path ./MOT17/images/train
    --data_txt_path_train ./datasets/data_path/mot17.half \
    --data_txt_path_val ./datasets/data_path/mot17.val \
    --resume model_final.pth \

```

--------------------------------

### Convert CrowdHuman to COCO Format

Source: https://github.com/foundationvision/bytetrack/blob/main/README.md

Convert the CrowdHuman dataset to COCO format using the provided script. Ensure you are in the ByteTrack home directory.

```shell
cd <ByteTrack_HOME>
python3 tools/convert_crowdhuman_to_coco.py
```

--------------------------------

### Compiler Flags

Source: https://github.com/foundationvision/bytetrack/blob/main/deploy/DeepStream/CMakeLists.txt

Sets C++ and CUDA compiler flags, including position-independent code generation.

```cmake
set(CMAKE_CXX_FLAGS "-Wall -Wextra -fPIC")
set(CMAKE_CUDA_FLAGS "--compiler-options -fPIC")
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.