Detectors¶
Detectors World has four detectors. Pick the one that matches what you want to find.
| Detector | Name | Finds | Results | Models download |
|---|---|---|---|---|
| Hand | "hand" |
Hands, 21 landmarks each, left or right | Landmarks |
7.8 MB |
| Pose | "pose" |
Bodies, 33 landmarks each | Landmarks |
5.8 to 30.7 MB |
| Face | "face" |
Face boxes and 6 key points | FaceBox |
0.2 MB |
| Face Mesh | "face_mesh" |
Faces, 478 landmarks each | Landmarks |
3.8 MB |
Which one should I use?¶
- Is someone there, and where? Use Face. It is the smallest model and returns simple boxes.
- Where are the eyes, lips or nose exactly? Use Face Mesh.
- What are the hands and fingers doing? Use Hand.
- How is the whole body moving? Use Pose.
Options every detector accepts¶
| Option | Default | Meaning |
|---|---|---|
detection_confidence |
0.5 (0.6 for "face") |
Minimum confidence, from 0.0 to 1.0, to report something new |
tracking_confidence |
0.5 |
Minimum confidence to keep following something between frames (not used by "face") |
static_image |
False |
Set to True for unrelated photos |
model_path |
None |
Use a model file you already have |
Options after these are specific to each detector and are listed on its page.