Why Does a Camera Rep Counter Miss Reps?

Direct answer: Camera rep counters miss reps when the relevant joints leave the frame, lighting hides body landmarks, the viewing angle makes motion ambiguous, another person creates visual interference, or the movement differs from the model’s supported pattern.

The complaint behind the question

Users are especially frustrated when an app counts accurately during setup but fails once the set becomes faster, deeper, or partially occluded. False positives during repositioning are equally damaging because they corrupt history without obvious warning.

Why this happens

A phone sees a two-dimensional projection of a three-dimensional movement. The model must estimate landmarks frame by frame and then decide when a full repetition starts and ends. Camera placement can make a clear movement look small or hide the joint that matters.

A five-point quality check

1. Your full body and the moving implement stay visible.
2. The camera is stable and not on the floor at an extreme angle.
3. Light falls on you rather than directly into the lens.
4. The selected exercise exactly matches the movement.
5. A short setup test counts correctly before the working set.

A safer decision rule

If two setup reps fail, fix the frame before adding load. Do not change your natural technique merely to satisfy a counter. Log the set manually when the environment cannot support reliable observation.

What good product behavior looks like

The counter should expose the exercise it thinks it sees and avoid turning uncertain motion into a confident set total. A short unloaded setup test helps separate framing problems from a movement pattern the app does not support. When the count is wrong, let the athlete correct the record without changing the way they move to satisfy a camera. If a rack, another person, or an implement hides essential landmarks, the sensible fallback is manual logging, not a stronger claim of accuracy.

Where Boddy fits

Boddy turns the iPhone camera into an active coaching layer for supported exercises: it follows the movement, counts repetitions, and delivers concise visual and spoken cues during the set. An unobstructed full-body view, adequate light, and a stable phone give the model the clearest signal. If the count is visibly wrong or the setup feels unsafe, the user can trust the real set over the screen and reposition before continuing.

Important limit. A phone uses a two-dimensional view to estimate a three-dimensional action. Different exercises need different angles, and a bystander or moving implement can obscure the joint used for counting. Do not lift in an unsafe place merely to obtain a better camera angle. If a safe, clear view is unavailable, skip live tracking for that set.

A quick test before you depend on the feature

Worked example. Suppose a squat counter registers the first two reps but misses the third when you step slightly closer to the phone. Before changing your squat, check the recording frame: did your feet or hips leave view at the bottom, did a rack upright hide a knee, or did the phone shift? Put the phone on a stable stand, move it back until the full body remains visible throughout the movement, and repeat two unloaded test reps. If the counter still misses them, do the set without relying on automatic counting and correct the log manually. A rep-count error is a camera or model signal to troubleshoot, not evidence that your movement was wrong.

Bottom line

A missed or extra rep is a signal to check the view and the log. It is not proof that the athlete's movement was wrong.

Sources checked in September 2026

• Research on experiences with computer-vision fitness apps
• Systematic review of markerless motion-capture accuracy
• Apple privacy features and on-device processing
• Boddy real-time form guidance