Science Competition

FreshScan — Visual Produce Quality Analyzer

Scan a fruit or vegetable and get an instant visual-quality read: freshness classification, damaged-area highlighting, shelf-life estimate, grading, and a suggested market for the produce.

What this actually measures: only what's visible in the photo — color, browning, dark or discolored patches, and pale/mold-like areas. It does not and cannot measure pH, pesticide/fertilizer residue, or internal chemical contamination — those require lab equipment (pH strips/meters, chromatography, spectrometry), not a camera. That limitation is intentional and disclosed in every result below.

New — Torch/Glare Coating Check: a flash-lit photo can reveal something a plain photo can't: whether the skin has an artificial wax/coating layer (common on apples, cucumbers, citrus). Wax reflects flash light as a large, smooth, mirror-like glare; natural skin scatters it into small, irregular highlights. This is still a visual proxy, not a chemical test — it tells you "this surface looks coated," not "this has pesticide X on it."

1 · Capture or upload

Original
No image yet — take or upload a photo above
Damage highlight
Highlighted map appears here after analysis

2 · Torch/glare coating check — optional

Turn your phone's flashlight on, hold it close to the same spot on the fruit (10–15cm), and snap a second photo. This checks for an artificial wax/coating layer by how the flash glare reflects off the skin — it does not detect chemicals or residue.

Torch photo
No torch photo yet (optional — skip if not needed)
Glare map
Glare map appears here after analysis

3 · Results

—
Affected area
—
Quality grade
—
Est. shelf life
—
Some spoilage — internal mold, bacterial growth, early decay — isn't visible from the outside. Camera-based screening supports a decision; it doesn't replace common sense or, for anything you're unsure about, a smell/texture check.
Suggested market
—
How the analysis actually works (methodology & limits)

The image is sampled pixel-by-pixel and converted from RGB to HSV color space. Background pixels (plain, low-saturation, bright — like a table or counter) are excluded. Among the remaining "produce" pixels, the app finds the dominant hue — the produce's characteristic healthy color — then flags a pixel as a defect if it is:

  • much darker than the item's typical brightness (bruising / rot), or
  • strongly shifted in hue from the dominant color (browning, decay spots), or
  • unusually pale and washed-out relative to the item (mold / dryness)

The percentage of flagged pixels becomes the "affected area." Grading (A/B/C) is a proxy based on this defect percentage and color consistency — it is not official USDA-style grading, which also weighs size and weight against a physical reference scale that a single photo can't provide. Shelf-life is a rough heuristic (typical days-to-spoilage for that produce type, scaled down by how much damage is already visible), not a lab-measured decay curve.

Torch/glare coating check (optional): when a flash-lit photo is provided, the app finds very bright, low-saturation "glare" pixels and measures two things: how much of the produce area they cover, and how smooth/contiguous the biggest glare blob is versus scattered into many small specks. Waxed or artificially coated produce tends to reflect flash light as one large, smooth, mirror-like patch because the coating fills in the skin's natural micro-texture. Unwaxed skin scatters the flash into many small, irregular highlights because of natural pores, ridges, and matte texture. This is a visual glossiness proxy for surface coating — it cannot identify what any coating or residue is made of, and it is not a pesticide/fertilizer/chemical test. A "High" reading means "this surface looks synthetically glossy," not "this fruit is unsafe."