Monocular_Distance_Velocity_Detect

Algorithm based on Yolo v5 and Deep Sort to detect surrounding vehicles' distance and relative velocity

Improved from Monocular_Distance_Detect

The Video Demo

HD Video can be found Here

Install

• vs2019 CUDA cuDNN: Make sure the version fit the requirement of your hardwares
• Anaconda(Recommend) Python>=3.8
• Pytorch: Check the version (MUST be GPU instead of CPU) of packages provided by conda before install
• pip install -r requirement.txt: install libraries before run the track.py script

Note: GPU NVIDIA 3060 and above should use pytorch>=1.11

Weights / Checkpoints

• Yolo_path: ./Monocular_Distance_Velocity_Detect/
• deep sort: C:\Users\Your_Computer_Name\.cache\torch\checkpoints\

if the program doesn't download the deep sort checkponts automatically, copy the files in /checkpoints to the correct path manually.

Run

python track.py --source YOUR_PATH\demo.mp4 --yolo_model yolov5m.pt --deep_sort_model osnet_x1_0_imagenet --show-vid --save-vid --save-csv

Note: yolov5m.pt & osnet_x1_0_imagenet could be selected by yourself.

Batch Run

if there are many videos under the same folder, modify and execute run.py

python run.py

Output

Output Path: ./runs/track/

Important Parameters in track.py

Video/Image Resolution

#Line 58
#Your video/image resolution/size
#画面分辨率
W = 1280
H = 720

Vertical Height

#vertical height(m) from camera to the ground/road
#相机离地面高度
H = 0.4

Angle

#The angle between the camera len and the horizontal line(the moving direction of vehicle), default is 0
#相机与水平线夹角, 默认为0 相机镜头正对前方，无倾斜
angle_a = 0

Detection Classes

In track.py, we only detect the ['person', 'car', 'truck', 'bicycle', 'motorcycle', 'bus'], follow this and modified the code(line 301, 325) to add more.

Output Path (saved file)

#please update the path before running the track.py script
if save_csv:
  df = pd.DataFrame(storage)
  df.to_excel('Your_path/test.xlsx',index=False)