curl --request GET \
--url https://api.hebbianrobotics.com/api/quality/cleanliness \
--header 'Authorization: Bearer <token>'import requests
url = "https://api.hebbianrobotics.com/api/quality/cleanliness"
headers = {"Authorization": "Bearer <token>"}
response = requests.get(url, headers=headers)
print(response.text)const options = {method: 'GET', headers: {Authorization: 'Bearer <token>'}};
fetch('https://api.hebbianrobotics.com/api/quality/cleanliness', options)
.then(res => res.json())
.then(res => console.log(res))
.catch(err => console.error(err));{
"analysis": "<string>",
"classification_version": "<string>",
"clips_analyzed": 1,
"clips_measured": 1,
"clips_without_identity": 1,
"clips_without_measurements": 1,
"dataset": "<string>",
"distorted_frames": 1,
"frame_quality": "<string>",
"frames_measured": 1,
"jerky_frames": 1,
"usable_frames": 1,
"distorted_share": 123,
"jerky_share": 123,
"usable_share": 123
}{
"code": "<string>",
"message": "<string>"
}{
"code": "<string>",
"message": "<string>"
}{
"code": "<string>",
"message": "<string>"
}Corpus-level cleanliness counts, shares and coverage reduced from stored per-frame measurements.
This endpoint performs no decode or model call. usable, distorted and
jerky form one exhaustive partition of frames_measured, stamped with the
opaque version of the rule that produced it.
curl --request GET \
--url https://api.hebbianrobotics.com/api/quality/cleanliness \
--header 'Authorization: Bearer <token>'import requests
url = "https://api.hebbianrobotics.com/api/quality/cleanliness"
headers = {"Authorization": "Bearer <token>"}
response = requests.get(url, headers=headers)
print(response.text)const options = {method: 'GET', headers: {Authorization: 'Bearer <token>'}};
fetch('https://api.hebbianrobotics.com/api/quality/cleanliness', options)
.then(res => res.json())
.then(res => console.log(res))
.catch(err => console.error(err));{
"analysis": "<string>",
"classification_version": "<string>",
"clips_analyzed": 1,
"clips_measured": 1,
"clips_without_identity": 1,
"clips_without_measurements": 1,
"dataset": "<string>",
"distorted_frames": 1,
"frame_quality": "<string>",
"frames_measured": 1,
"jerky_frames": 1,
"usable_frames": 1,
"distorted_share": 123,
"jerky_share": 123,
"usable_share": 123
}{
"code": "<string>",
"message": "<string>"
}{
"code": "<string>",
"message": "<string>"
}{
"code": "<string>",
"message": "<string>"
}Authorizations
Capability-bearing token issued by a squash operator. Required for curation, exports, dataset registration, and job control when the deployment runs in gated mode; public reads stay open. GET /api/access reports the capabilities a given credential actually carries.
Query Parameters
Dataset or corpus id whose index root holds the measurements.
Stored analysis whose clips define the expected population. Omit for the most recent.
Stored raw frame-measurement run to classify. Omit for the most recent.
Response
Stored frame-quality counts, shares and coverage
Response of GET /api/quality/cleanliness: a corpus-level frame partition
reduced from one stored analysis and one stored frame-quality run.
Counts are the source of truth and shares are conveniences derived from the
same frames_measured denominator. Shares are absent when no frame was
measured, which is different from measuring frames and finding zero in a
category.
The cutoffs that produced the partition are NOT published. They are the
calibration of our measuring instrument, which the contract reserves to us,
and a reader who could vary them could re-partition the corpus to whatever
answer suited them. classification_version is what makes two partitions
comparable, which is the only thing the cutoffs would have told them.
The run whose clip identities defined the expected population — the one asked for, or the most recent when none was named.
Opaque token naming the classification rule. Two partitions carrying the same token are directly comparable; two carrying different ones are not.
x >= 0x >= 0Legacy analysis artifacts that cannot be joined because they recorded no stable source identity.
x >= 0Identifiable analysed clips with no object in the measurement run.
x >= 0x >= 0The stored frame-measurement run that was classified, resolved the same way.
x >= 0x >= 0Frames free of the distortion and jerk conditions below. Internally the
historical category name is active; this wire name avoids confusion
with active manipulation.
x >= 0