Define a Parking-Space Decision
The owner asks for a trustworthy count of free spaces.
Read the full lesson and open its transcript below. A narrated video will be added after production.
What you will learn
The owner asks for a trustworthy count of free spaces.
A practical question
The garage operator asks, 'Is this space free?' Define the answer before writing code. Does a motorcycle occupy it? What if a vehicle crosses the space without parking? How long must the state persist before the sign changes?
The useful idea
Mark each parking-space region in a calibrated camera view. Define occupied, empty, and uncertain states with examples. Choose an evaluation unit such as a space at a particular time, and identify when a human should review the image.
Work through the example
Walk through a clear empty space, a parked car, and a partly hidden motorcycle with the operator. If the policy is ambiguous, resolve it here rather than hiding it inside a model threshold. A car rolling across a parking space is visible but has not necessarily parked there. Define whether occupancy begins when the car stops, crosses a line, or remains for a minimum interval. Test the rule on a passing vehicle and on a genuine arrival. If the view is blocked, mark uncertainty rather than inventing a state. Write the policy beside the camera geometry so annotators and the deployed system count the same event. Use this case to test your understanding: Specify what to do when a moving vehicle crosses a space without parking. Here is the limitation to watch for: Without a decision rule, two correct detectors can give different counts. The answer follows from that evidence: Define one occupancy decision per marked region, its time window, and rules for uncertain evidence. Test a car passing through the space and one that parks. If the decision rule cannot distinguish them, define a waiting interval or uncertain state before using the count. Keep this event rule consistent in labels and production. A rule that changes between the two makes evaluation misleading.
Limits and next step
Next, we'll connect capture, geometry, detection, decision, and logging into one reproducible pipeline.
Worked example
Specify what to do when a moving vehicle crosses a space without parking.
Quick check: test your understanding
Try it yourself: Consider this case: Specify what to do when a moving vehicle crosses a space without parking. Explain how you would check the result. In your answer, address this limitation: Without a decision rule, two correct detectors can give different counts.
Check the reasoning
Define one occupancy decision per marked region, its time window, and rules for uncertain evidence. Without a decision rule, two correct detectors can give different counts.
What is the main idea behind define a parking-space decision?
Define one occupancy decision per marked region, its time window, and rules for uncertain evidence.
How is it used in the example?
Specify what to do when a moving vehicle crosses a space without parking.
What limitation should be checked?
Without a decision rule, two correct detectors can give different counts.
Further reading
Programming Computer Vision with Python — Jan Erik Solem.
Original ToolsInVision teaching examples. Editorial approach · Suggest a correction
Accessible referenceLesson 77 transcript302 words · Narration ready for the upcoming videoOpen transcript
The garage operator asks, 'Is this space free?' Define the answer before writing code. Does a motorcycle occupy it? What if a vehicle crosses the space without parking? How long must the state persist before the sign changes?
Mark each parking-space region in a calibrated camera view. Define occupied, empty, and uncertain states with examples. Choose an evaluation unit such as a space at a particular time, and identify when a human should review the image.
Walk through a clear empty space, a parked car, and a partly hidden motorcycle with the operator. If the policy is ambiguous, resolve it here rather than hiding it inside a model threshold. A car rolling across a parking space is visible but has not necessarily parked there. Define whether occupancy begins when the car stops, crosses a line, or remains for a minimum interval. Test the rule on a passing vehicle and on a genuine arrival. If the view is blocked, mark uncertainty rather than inventing a state. Write the policy beside the camera geometry so annotators and the deployed system count the same event. Use this case to test your understanding: Specify what to do when a moving vehicle crosses a space without parking. Here is the limitation to watch for: Without a decision rule, two correct detectors can give different counts. The answer follows from that evidence: Define one occupancy decision per marked region, its time window, and rules for uncertain evidence. Test a car passing through the space and one that parks. If the decision rule cannot distinguish them, define a waiting interval or uncertain state before using the count. Keep this event rule consistent in labels and production. A rule that changes between the two makes evaluation misleading.
Next, we'll connect capture, geometry, detection, decision, and logging into one reproducible pipeline.