Access the Inside the Fridge sample
Access is granted immediately after you submit this form. Your Hugging Face username and email are shared with CollectData for access administration and dataset notices.
I have read and accept the CollectData Sample Data License 1.1 linked above. If accepting for an organization, I have authority to bind it.
Log in or Sign Up to review the conditions and access this dataset content.
Inside the Fridge
Explore the contents of 44 real household refrigerators in 47 trimmed, redacted videos, with one wide-view thumbnail per fridge. About 41 minutes of video capture shelves, packaging, food, and everyday clutter from changing viewpoints.
Use this free sample to explore food and packaging recognition, video understanding, household scene understanding, and perception for robotics. Research and commercial model development are permitted under the CollectData Sample Data License 1.1.
About CollectData · Contact CollectData for a collection tailored to your use case.
Browse the fridges
These small JPEG previews are visible to everyone, without signing in. Select a fridge to browse its files. Full-resolution videos require access approval above; approval is automatic after you accept the license.
Get the data
Select Request access, complete the short form, and accept the license. Access is approved automatically, as with our Tidy Up sample. No separate email request is needed.
Download individual files from Files and versions, or download this release after accepting access:
pip install -U huggingface_hub
hf auth login
hf download CollectDataIO/inside-the-fridge --repo-type dataset --revision v1.0.0 --local-dir inside-the-fridge
Each fridges/fridge-XX/ directory contains thumbnail.webp and one or more part-XX.mp4 files. Public fridge numbers match the gallery and have intentional gaps. assets.csv has one row per video; manifest.json lists all 91 media files, their sizes, checksums, and video properties. SHA256SUMS covers the release files.
What is included
| Item | Contents |
|---|---|
| Refrigerators | 44 |
| Videos | 47 MP4 clips; 41.3 minutes total |
| Thumbnails | 44 WebP images; one wide view per fridge |
| Encoding | HEVC/H.265, 30 fps, retained display resolution (see manifest) |
| Audio | None: every MP4 contains exactly one video track and zero audio tracks |
| Annotations | File-level metadata only; no food labels or bounding boxes |
The three multipart fridges each have two retained clips. The website joins these parts and uses smaller H.264 versions for browser playback. This download retains the higher-resolution reviewed HEVC videos. Use an HEVC-capable player or decoder.
Preparation and limitations
This release contains selected open-fridge intervals from paid contributions. Clips were trimmed, reviewed for faces and visible personal information, and masked where issues were found. Two tattoo-bearing scans were excluded. All known retained review findings are masked. The review combined automated face detection with human review and sampled checks; it does not certify complete anonymization. Do not attempt to identify contributors or their homes. Report privacy issues to info@collectdata.io.
The release excludes raw captures, participant identifiers, GPS, private review evidence, and unreviewed ancillary files. It does not include depth maps, poses, meshes, or synchronized scan bundles. Clip times are normalized and are not timestamps into raw captures. Coverage, framing, lighting, occlusion, and visible contents vary; this is a small convenience sample, not a representative food inventory benchmark.
License and attribution
The unchanged CollectData Sample Data License 1.1 governs this release; license.json binds it to the package manifest. There is no dataset license fee, revenue threshold, per-model royalty, or scheduled expiry. The license permits research and commercial model development and deployment, subject to its terms. You may publish compliant results, models, and reconstructions. Credit CollectData when sharing results, protect privacy, and do not repost or redistribute the source dataset. Share this Hugging Face page so others can accept their own license.
Suggested credit: Created using the CollectData Inside the Fridge dataset — https://huggingface.co/datasets/CollectDataIO/inside-the-fridge.
@misc{collectdata2026fridge,
author = {{CollectData}},
title = {Inside the Fridge: Real Household Refrigerator Videos},
year = {2026},
version = {1.0.0},
publisher = {Hugging Face},
url = {https://huggingface.co/datasets/CollectDataIO/inside-the-fridge},
note = {Release tag v1.0.0}
}
- Downloads last month
- 28