Get Per Sample Report
curl --request GET \
--url https://api.landing.ai/v1/projects/{project_id}/train/{training_id}/results/per-sample-reportimport requests
url = "https://api.landing.ai/v1/projects/{project_id}/train/{training_id}/results/per-sample-report"
response = requests.get(url)
print(response.text)const options = {method: 'GET'};
fetch('https://api.landing.ai/v1/projects/{project_id}/train/{training_id}/results/per-sample-report', options)
.then(res => res.json())
.then(res => console.log(res))
.catch(err => console.error(err));{
"data": {
"downloadExpiresAt": "<string>",
"downloadUrl": "<string>"
}
}{
"detail": [
{
"loc": [
"<string>"
],
"msg": "<string>",
"type": "<string>"
}
]
}Training
Get Per Sample Report
Get a download link for the per-sample performance report of a model.
The report is a CSV file with one row per image in the snapshot the model was trained on, covering every split, and it is the same file the LandingLens web app offers as a model performance CSV download.
The columns are Project Name, Project Type, Image Name, Image ID,
Split, Upload Time, Image Width, Image Height, Model Name,
Metadata, followed by the ground truth and prediction columns for the
project type:
- Classification:
GT_Class,PRED_Class,Model_Correct,PRED_Confidence,Gt-PRED JSON - Anomaly detection:
GT_Class,PRED_Class,Model_Correct,Model_Confidence,Gt-PRED JSON - Object detection:
GT_Class,PRED_Class,PRED_Class_Confidence,Class_TotalArea,GT-PRED JSON
Predictions are reported at the threshold saved on the model, which is the
same threshold used by the results/metrics and results/confusion-matrix
endpoints. Segmentation projects are not supported.
The download link expires; downloadExpiresAt is when it does.
GET
/
v1
/
projects
/
{project_id}
/
train
/
{training_id}
/
results
/
per-sample-report
Get Per Sample Report
curl --request GET \
--url https://api.landing.ai/v1/projects/{project_id}/train/{training_id}/results/per-sample-reportimport requests
url = "https://api.landing.ai/v1/projects/{project_id}/train/{training_id}/results/per-sample-report"
response = requests.get(url)
print(response.text)const options = {method: 'GET'};
fetch('https://api.landing.ai/v1/projects/{project_id}/train/{training_id}/results/per-sample-report', options)
.then(res => res.json())
.then(res => console.log(res))
.catch(err => console.error(err));{
"data": {
"downloadExpiresAt": "<string>",
"downloadUrl": "<string>"
}
}{
"detail": [
{
"loc": [
"<string>"
],
"msg": "<string>",
"type": "<string>"
}
]
}