Details on request and response JSON for 3D box pre-labeling model API
Deepen AI uses a trained model for predicting labels for 3D boxes. We have created the API for detecting boxes in an image. The details are as follows:
Endpoint: https://tools.deepen.ai/api/v2/datasets/{dataset_id}/3d_bbox_inference
Content-type : none
Request JSON: none
Response JSON data :
{
"labels":{
"00000.pcd":{
"Vehicle:1":{
"label_category_id":"Vehicle",
"label_id":"Vehicle:1",
"score":0.36743032932281494,
"three_d_bbox":{
"w":1.8980461359024048,
"h":1.6315248012542725,
"l":4.4147796630859375,
"vx":-0.16655054688453674,
"vy":4.0891852378845215,
"cx":-3.8604888916015625,
"cy":19.579147338867188,
"cz":0.9247932844882278,
"rot_z":1.5279546976089478
}
}
}
}
}
The following 3 classes can be predicted for boxes:
VEHICLE, PEDESTRIAN, CYCLIST