For each food it recognizes, MorselFlow looks for one of three outcomes: a count of clearly separate pieces, a small/typical/large comparison with a standard serving, or “unknown” when the photo does not provide enough evidence.
Three possible estimates
MorselFlow changes its approach according to what the photo shows.
Count
Count visible pieces
For dumplings, sushi, rice balls, or bread, MorselFlow can count the pieces that are fully visible and clearly separate. Hidden pieces are not observable.
Relative size
Compare with a typical serving
For rice, noodles, or curry, MorselFlow compares the visible serving with its standard serving and suggests small, typical, or large.
Unknown
Know when it is unclear
When food is hidden, overlapping, partly eaten, or missing a useful size cue, MorselFlow can leave the visual portion uncertain instead of claiming a precise amount.
An estimate, not a measurement
A single photo cannot measure weight or reveal hidden food.
The image does not provide the exact dimensions of a bowl, the density of the food, or the amount hidden underneath. Two servings can occupy a similar area in a photo but weigh different amounts.
MorselFlow combines the visual estimate with its own standard portion data. The resulting grams and nutrition values remain estimates; the image model does not directly measure either one.
When to check more carefully
Visible boundaries improve the draft; occlusion weakens it.
More usable visual evidence
- The whole meal is visible in the frame
- Individual pieces are not hidden or overlapping
- A plate or bowl provides a useful size cue
- The photo was taken before eating
Less usable visual evidence
- Part of the dish is outside the frame
- Foods overlap and their edges are unclear
- Soup or sauce hides much of the portion
- The meal has already been partly eaten
A 10-second photo check
Show the full serving before you eat.
Keep the full dish in frame
Include the rim of the plate or bowl. The container gives MorselFlow a better visual reference than a close crop of the food alone.
Take the photo before eating
The original serving is more informative than trying to reconstruct it from what remains.
Do not rely on overhead photos alone
A slight angle may show food height or bowl depth that an overhead view hides. Use the angle that keeps every dish visible.
How MorselFlow uses it
Check the food, piece count, and serving size before saving.
After you choose a photo, MorselFlow suggests likely foods and amounts. Check whether it identified the right dish, counted every visible piece, and chose a plausible serving size.
If a suggestion is wrong, change it before saving. If exact grams matter—for example, for a recipe or clinical instruction—use a scale or the information supplied with the food.
The feature is designed to reduce initial typing, not to replace measurement.
Food candidates, portions, and nutritional values are estimates based on photos and other available information. They may differ from the actual meal. MorselFlow is not intended for medical diagnosis or treatment.