A tool to visualize the trajectory of drones in videos

If you have better ideas, welcome to propose them!! Please kindly star ⭐ this project if it helps you

Modes

This project now keeps both implementations:

• legacy: the original motion-overlay method based on frame differencing against the first frame
• improved: an optional hybrid mode that keeps the original overlay effect and adds a smoothed tracked trajectory line on top

The original implementation still exists and is not replaced. The improved result is only a selectable mode.

GUI Highlights

• GUI now supports explicit mode switching between legacy and improved
• improved mode exposes selectable line colors, line thickness, and glow strength
• GUI shows status and failure messages directly inside the window instead of relying on terminal output
• built-in recommended parameter presets are available for both modes

Still shot videos:

And the result:

Although sometimes it can be wrong, this tool works well in certain circumstances. If you are looking for higher quality images, you may need to do it manually!

Update

1. Added a UI interface for more intuitive display

2. Optimized the generation logic

   PS: the video is come from composite_image

   

3. Add more parameters and set gradient transparency overlay

4. Added an optional improved visualization mode:

   • preserve the original legacy output
   • track the drone trajectory with a lightweight rule-based tracker
   • smooth the trajectory and draw a clean highlight line on top of the original overlay
   • support CLI export for legacy, improved, and comparison outputs

Improved Example

Improved hybrid output:

Legacy vs improved comparison:

Parameter Description

采样间隔（sample_interval_entry）

作用：每隔多少帧进行一次处理（间隔越大，计算越快，但可能导致轨迹不连续）。 默认值：10（每10帧进行一次分析）。

差分阈值（diff_threshold_entry）

作用：用于二值化处理，设置像素差值的阈值（数值越低，检测的运动更灵敏）。 默认值：30（像素差大于30的部分视为运动）。

膨胀核大小（kernel_size_entry）

作用：对检测到的运动区域进行膨胀，以去除噪声（数值越大，运动区域轮廓越平滑）。 默认值：15（15×15的核进行膨胀）。

起始透明度（start_alpha_entry）

作用：决定运动区域最初的透明度（0.0 完全透明，1.0 不透明）。 默认值：0.2（运动区域初始时 20% 透明）。

最终透明度（end_alpha_entry）

作用：控制运动区域的最终透明度（在处理过程中逐渐增加）。 默认值：1.0（最终完全不透明）。

Recommend

This project is also very good

Disadvantages

Manual parameter adjustment

Please manually select the appropriate parameters according to your own video

Handling different speeds

If the speed of the drone changes in the video, the trajectory may not be good.

The influence of background

If the background moves, such as lighting changes or pedestrian movement, the superposition may be incorrect.

CLI Usage

Export the original implementation (legacy):

python3 visualize_uav_trajectory.py \
  --video example_video.mp4 \
  --method legacy \
  --output images/example_legacy.png

Export the optional improved result:

python3 visualize_uav_trajectory.py \
  --video example_video.mp4 \
  --method improved \
  --output images/example_improved.png

Export a side-by-side comparison image:

python3 visualize_uav_trajectory.py \
  --video example_video.mp4 \
  --compare \
  --output images/example_comparison.png

If you omit --video, the script starts the Tkinter GUI. The original implementation remains available in code as the legacy mode, and the improved mode is an extra option for CLI export and further extension.

Note: the images/ directory is ignored by git. For repository examples shown in this README, the exported result images are copied into example/ before commit.

Project Structure

The code is split to make future maintenance easier:

• visualize_uav_trajectory.py: thin entry point
• uav_vis/config.py: shared constants and style definitions
• uav_vis/core.py: legacy rendering, improved tracking, and image generation logic
• uav_vis/gui.py: Tkinter GUI and in-window status handling
• uav_vis/cli.py: CLI parser and command execution flow