Frequently Asked Questions
Everything you need to know about getting the best results from Param's video analysis.
Processing time depends on the length of your video. Here's what to expect:
- Short clips (1-5 minutes): Typically 5-15 minutes to process
- Full matches (30-60 minutes): Can take 45 minutes to 2 hours
You'll receive real-time progress updates on your dashboard, and we'll notify you when processing is complete. The system uses GPU-accelerated processing to ensure the fastest possible results.
Tip: Processing happens in the cloud, so you can close your browser and come back later. Your results will be waiting for you.
Param offers two processing modes that use different AI models under the hood:
Fast Mode
- Uses YOLO for detection
- Significantly faster processing
- Great accuracy for most use cases
- Recommended for most users
Slow Mode
- Uses SAM3 for segmentation
- Longer processing time
- Slightly more precise tracking
- For when you need exact data
Which should I choose?
For the vast majority of users, Fast mode is the best choice. The YOLO model provides excellent accuracy that's more than sufficient for match analysis, movement tracking, and performance insights.
Slow mode uses SAM3 (Segment Anything Model 3), which provides marginally better precision at the cost of significantly longer processing times. The difference in accuracy is minimal for typical coaching and review purposes.
Recommendation: Use Fast mode unless you specifically need extremely precise positional data for research or advanced biomechanical analysis.
For the most accurate analysis, your videos should meet these criteria:
Use a Static Camera
The camera should be mounted or placed on a stable surface and not move during recording. Moving cameras make it difficult for our AI to accurately track player positions and movements.
Show the Entire Court
The video should capture the full squash court in frame at all times. This allows us to:
- Track both players throughout the rally
- Calculate accurate court coverage and positioning
- Measure movement patterns and distances traveled
- Analyze the T-position control and court dominance
Recommended Setup
- Position: Behind the back wall glass, centered on the court
- Height: Elevated position (balcony or high tripod) for a clear overhead-ish view
- Resolution: 720p minimum, 1080p recommended
- Frame rate: 30fps or higher
Avoid: Handheld recordings, videos that zoom in/out, or angles that only show part of the court. These will result in incomplete or inaccurate analysis.
Reference points are markers you set on your video to help our system understand the court geometry. They're essential for converting pixel positions into real-world court coordinates.
Why Reference Points Are Important
Every court is filmed from a different angle and distance. Reference points allow us to:
- Map the court accurately: Transform the camera's perspective into a standardized court view
- Calculate real distances: Know exactly how far players move in meters, not just pixels
- Generate heatmaps: Show precise positioning on a normalized court diagram
- Compare across videos: Analyze performance consistently regardless of camera setup
Tip: Take your time setting reference points accurately. A few seconds of careful clicking will significantly improve the quality of your analysis results.