Move a Camera with Extrinsics
A technician moves the parking camera to a new pole.
Move a Camera with Extrinsics explained visually
A technician moves the parking camera to a new pole. English captions are optional and start off. The complete transcript follows the written lesson.
What you will learn
A technician moves the parking camera to a new pole.
A practical question
We move the parking camera from one pole to another. The camera's internal settings might remain the same, but every vehicle now appears in a different part of the frame. Extrinsics describe that change in viewpoint.
The useful idea
They give a rotation and translation relating a named world frame to a named camera frame. Pay attention to the direction: a world-to-camera transform is not the same as a camera-to-world transform. Once a world point is expressed relative to the camera, the internal camera model can project it onto the image.
Work through the example
Choose a painted corner whose map position is known. Try the old camera pose and the new one, then compare both projected positions with the photos. A consistent offset or flipped direction is a cue to revisit the transform convention, not to nudge every individual parking bay by hand.
Limits and next step
Save the pole location, pose definition, and image. Next, we'll verify the full projection with measured image points rather than trusting the calibration numbers alone.
Worked example
Transform a fixed painted corner into each camera's coordinate frame before projecting it.
Quick check: test your understanding
Try it yourself: In the situation described above, explain the method and one reason its output might be misleading.
Check the reasoning
Extrinsics describe the rotation and translation between a world frame and camera frame. Changing pose invalidates image regions drawn for the old position.
What is the main idea behind move a camera with extrinsics?
Extrinsics describe the rotation and translation between a world frame and camera frame.
How is it used in the example?
Transform a fixed painted corner into each camera's coordinate frame before projecting it.
What limitation should be checked?
Changing pose invalidates image regions drawn for the old position.
Further reading
Programming Computer Vision with Python — Jan Erik Solem.
Original ToolsInVision teaching examples. Editorial approach · Suggest a correction
Accessible referenceLesson 19 transcript162 words · Video transcript · Matches the published narration and captionsOpen transcript
We move the parking camera from one pole to another. The camera's internal settings might remain the same, but every vehicle now appears in a different part of the frame. Extrinsics describe that change in viewpoint.
They give a rotation and translation relating a named world frame to a named camera frame. Pay attention to the direction: a world-to-camera transform is not the same as a camera-to-world transform. Once a world point is expressed relative to the camera, the internal camera model can project it onto the image.
Choose a painted corner whose map position is known. Try the old camera pose and the new one, then compare both projected positions with the photos. A consistent offset or flipped direction is a cue to revisit the transform convention, not to nudge every individual parking bay by hand.
Save the pole location, pose definition, and image. Next, we'll verify the full projection with measured image points rather than trusting the calibration numbers alone.