Create a photo dataset of cucumbers in trays (multiple objects per image) for QC / computer vision

Job description:

I’m looking for a freelancer to take a set of photos of cucumbers arranged white background, with multiple cucumbers per image, for a prototype computer vision project (quality control of sorting). No programming is required. The task is purely data collection (photos) following clear guidelines.

What should be in the photos

- A tray/crate containing several to a dozen+ cucumbers per image

- Different arrangements: loose, dense, partly touching/overlapping, varied orientations

- “Mostly one type” trays + mixed trays (e.g., mostly straight but with a few curved/damaged)

Shooting conditions (important)

- Repeatable setup: same place, similar lighting, consistent camera angle

- Preferably top-down or slightly angled top-down (consistent across the dataset)

- Avoid harsh shadows and reflections

- Tray background should be relatively uniform and ideally contrasting with cucumbers (e.g., light tray)

- Images must be sharp (no motion blur)

- No beauty filters / AI enhancement / heavy post-processing

File requirements

- Format: JPG or PNG

- Resolution: minimum 1920×1080 (higher is better)

- Deliver original files (not re-compressed by messaging apps)

Dataset size

- Target: 200–400 images (can be delivered in phases)

- I prefer starting with a pilot batch (5–20 images) for quality check, then proceed with the full set.

Simple labeling

No bounding boxes or masks required. Just organize files into folders like:

- tray_straight_majority

- tray_curved_majority

- tray_mixed

- tray_damaged_present (optional, if available)

Delivery

- ZIP archive + link to download

- Include a short note describing the setup (camera height, angle, lighting)

Please include in your offer

- Your device/camera model (smartphone is fine)

- Whether you can keep a fixed camera setup (tripod/stand)

- How many images you can deliver