Skip to page content
SAVRN
Search Plan a project

Dataset · Image segmentation

cd-ogi-recording-mask-audit

by Sankalp Thakur sankalpsthakur/cd-ogi-recording-mask-audit

Audit tables for a small gas-imaging study on baselab's Controlled Diet dataset, which is laboratory methane from rumen fermentation. Its 137 rows describe 42 source images from nine recordings. They are not images, chemical measurements or gas-free sessions.

Rows137
Configurations3
Size151.4 KB
Licensecc-by-4.0
AccessPublicly accessible
Monthly Downloads52

Dataset Card

By Sankalp Thakur, published under cc-by-4.0, revision 7e336e3ce309.

Controlled Diet recording/mask audit

Audit tables for a small gas-imaging study on baselab's Controlled Diet dataset, which is laboratory methane from rumen fermentation. Its 137 rows describe 42 source images from nine recordings. They are not images, chemical measurements or gas-free sessions.

Config Split Rows Contents
recording_manifest development_train / _tune / _compare 18 / 12 / 12 Filenames, recording IDs, source hashes, diet class IDs, publisher foreground counts
compare_frames development_compare 84 12 frames × 7 models: native 640×480 confusion counts, IoU, Dice, precision, recall
candidates development_tune 11 Every candidate's outcome, selection flags and tuning IoU

The spatial prior's comparison IoU is 0.138078, against 0.026416 for the tiny MLP and 0.022555 for logistic regression: both learned models lose. Empty cells are undefined metrics, not zeros. background_pixel_overprediction_ratio measures pixel overprediction against weak masks. It is never a gas false-alarm rate.

Models, weights and the full report are in controlled-diet-ogi-mask-reference.

Source, attribution and modifications

Source: Controlled Diet (CD) Dataset for Methane Plume Detection, by Mohamed G. Embaby, Toqi Tahamid Sarker, Amer AbuGhazaleh and Khaled R. Ahmed, Southern Illinois University Carbondale. Exact revision 9475f96c32dd4764471e4358ca6659b0dd348ecd. Pinned source license declares CC BY 4.0; a source metadata JSON separately declares CC0. The unresolved discrepancy is retained; this audit elects CC BY 4.0 and makes no public-domain claim. See ATTRIBUTION.md, LICENSE and CC BY 4.0 terms. No endorsement is implied.

Changes: prospectively selected 42 pairs; recording-disjoint development roles; nonzero binary mask occupancy; fractional training targets and native-resolution confusion counts; CSV transfer of the frozen source IDs and CPU results. Filenames, source hashes, bytes and publisher counts are retained. No copied article or image payload is in this audit repository. The related model's original cloud archive retains the qualified PNG subset with its source notices and separate dependency grants.

The previous card, with selection hashes and publication records, is in HISTORY.md.

Structure

candidates 11 rows

SplitRowsSize
development_tune115.5 KB
candidate_idstringarmstringstatusstringselected_before_comparebooll2float64steps_completedint64objective_initialfloat64objective_finalfloat64objective_trace_countint64objective_increasesint64selected_tune_thresholdfloat64tune_recording_macro_ioufloat64weights_sha256stringoutcome_sha256stringtune_raw_probability_sha256stringsource_revisionstringselection_sha256stringprepared_seal_sha256stringconfirmation_allowedbool

compare_frames 84 rows

SplitRowsSize
development_compare8462.7 KB
armstringcandidate_idstringthresholdfloat64filenamestringrecording_groupstringdevelopment_rolestringdiet_class_idint64tpint64fpint64fnint64tnint64reference_foreground_pixelsint64predicted_foreground_pixelsint64ioufloat64dicefloat64precisionfloat64recallfloat64background_pixel_overprediction_ratiofloat64source_reported_foreground_pixelsint64image_source_pathstringimage_source_bytesint64image_source_sha256stringmask_source_pathstringmask_source_bytesint64mask_source_sha256stringsource_datasetstringsource_revisionstringselection_sha256stringweights_sha256stringprepared_seal_sha256stringconfirmation_allowedboolfield_qualifiedbool

recording_manifest 42 rows

SplitRowsSize
development_train188.8 KB
development_tune125.9 KB
development_compare125.9 KB
filenamestringrecording_groupstringdevelopment_rolestringoriginal_publisher_splitstringframe_indexint64diet_class_idint64source_reported_foreground_pixelsint64image_source_pathstringimage_source_bytesint64image_source_sha256stringmask_source_pathstringmask_source_bytesint64mask_source_sha256stringsource_datasetstringsource_revisionstringselection_sha256stringconfirmation_allowedboolphysical_release_independence_verifiedbool

Details

Repository
sankalpsthakur/cd-ogi-recording-mask-audit
Publisher
Sankalp Thakur
Task category
Image segmentation
Tags
optical-gas-imaging, audit, weak-masks
Size category
Not stated by the source
Languages
en
Revision
7e336e3ce3094dd11c74aae5f213964575c58176
Last updated
2026-10-07

Files

13 files, 151.4 KB in total.

Data8 files · 137.3 KB
Documentation4 files · 11.6 KB
Repository1 file · 2.5 KB
Every file
FileTypeSizeSHA-256
RELEASE-MANIFEST.jsonData3.6 KB—
candidates.csvData5.6 KB—
compare_frames.csvData61.1 KB—
protocol.jsonData4.1 KB—
recording_manifest/development_compare.csvData5.8 KB—
recording_manifest/development_train.csvData8.5 KB—
recording_manifest/development_tune.csvData5.7 KB—
recording_selection.jsonData43.0 KB—
ATTRIBUTION.mdDocumentation2.4 KB—
HISTORY.mdDocumentation4.9 KB—
LICENSEDocumentation976 B—
README.mdDocumentation3.3 KB—
.gitattributesRepository2.5 KB—

License and Download

License
cc-by-4.0
Access
No access gate
Download from Sankalp Thakur

Released by Sankalp Thakur through its official repository on Hugging Face. Read the license.

Models Trained on This Dataset