This application is meant to detect people from a live or recorded video stream and determine if they are entering or exiting a location. That info is then can be sent via a POST call to a URL endpoint for processing.
ENTRY-EXIT ZONES The app uses configurable entry and exit zones to determine if someone is entering or exiting a location. Zones are made up of rectangular areas in the video frame. Multiple entry and exit zones may be added to fit complex environments.
ENTER CRITERIA: When a detected person crosses from an entry to an exit zone, then disappears from tracking.
EXIT CRITERIA: When a detected person crosses from exit to an entry zone, then disappears from tracking.
IGNORE CRITERIA: Any detected person who does not cross both an entry and exit zone before tracking of them ends.
JSON PAYLOAD
When either an entry or exit is detected, the app will send a call to the URL endpoints designated in the config.json file. The payload consists of the following:
{
"device_id": "<id_of_app_instance>",
"timestamp": <UTC seconds epoch>
}
- Sign up for alwaysAI
- Install alwaysAI tooling
This app can run on either a desktop/laptop development machine or on an edge device such as a Jetson Nano, Dragonboard, or Raspberry Pi that has docker installed. See the above install instruction link for more details.
Edit the config.json file to adjust the following parameters:
| PARAMETER | DESCRIPTION | DEFAULT | REQUIRED |
|---|---|---|---|
| camera_id | Id of camera to use for camera detection. 0 if only one camera attached. 0 is usually the id for a built-in web cam, and 1 or higher if usb external cams are used in addition to an embedded camera | 0 | No |
| confidence | Threshold for accepting a detection as valid | 0.5 | No |
| enable_logs | Should print to console additional info | True | No |
| entry_zones | List of dictionaries defining bounding zones for parts of a frame where targets will be entering from | None | Yes |
| exit_zones | List of dictionaries defining bounding zones for parts of the frame where targets exiting a location will be last seen | None | Yes |
| device_id | Optional identifier for a different instance of the detector for multi-cam set ups | Device MAC Address | No |
| deregister_frames | Number of frames to count past last seen before a centroid tracker should consider an object lost | 50 | No |
| enable_streamer | Enable streamer to display vision output to localhost:5000 |
false | No |
| filename | Optional filename path for a file to use with the mode:File option |
None | No |
| max_distance | Maximum distance in pixels that a centroid tracker will search to maintain tracking on an object | 50 | No |
| mode | Camera or File supported. When using File, you must also provide a filename value |
Camera | No |
| model_id | Model id to use for detecting target labels. If only 1 model specified in the alwaysai.app.json file, models object, then this value should match |
None | No |
| server_entry_url | Endpoint for processing entry event | None | Yes |
| server_exit_url | Endpoint for processing exit event | None | Yes |
| target_labels | List of string labels that the model should filter for. If not provided than object detector will not filter | ["Person"] | No |
The following are different examples of entry and exit zone configurations that can be used in the config.json file.
For a camera placed perpendicular to traffic with a resolution of 1280x720:
"entry_zones": [
{
"threshold": 0.2,
"start_x": 0,
"start_y": 0,
"end_x": 300,
"end_y": 720
}
],
"exit_zones": [
{
"threshold": 0.2,
"start_x": 1000,
"start_y": 0,
"end_x": 1280,
"end_y": 720
}
]
For a camera placed to the side and looking down a curved hallway, for a frame resolution of 1280x720:
"entry_zones": [
{
"threshold": 0.2,
"start_x": 400,
"start_y": 200,
"end_x": 650,
"end_y": 400
}
],
"exit_zones": [
{
"threshold": 0.2,
"start_x": 1000,
"start_y": 0,
"end_x": 1280,
"end_y": 720
}
]
For a camera looking straight down a walkway with a resolution of 1280x720:
"entry_zones": [
{
"threshold": 0.2,
"start_x": 0,
"start_y": 0,
"end_x": 100,
"end_y": 720
},
{
"threshold": 0.2,
"start_x": 100,
"start_y": 0,
"end_x": 1280,
"end_y": 100
},
{
"threshold": 0.2,
"start_x": 100,
"start_y": 620,
"end_x": 1280,
"end_y": 720
},
{
"threshold": 0.2,
"start_x": 1180,
"start_y": 100,
"end_x": 1280,
"end_y": 620
}
],
"exit_zones": [
{
"threshold": 0.2,
"start_x": 540,
"start_y": 200,
"end_x": 740,
"end_y": 520
}
]
Frame x,y beging with 0,0 in the upper-left hand corner of the frame.
Docs: https://dashboard.alwaysai.co/docs/getting_started/introduction.html
Community Discord: https://discord.gg/alwaysai
Email: [email protected]