aicamcam.github.io

(Project added to TensorFlow.js gallery > Demos & Applications)

• Webcam video recrding when object is detected
• Object(person/dog/cat) detection and recording video
• Running Locally without any server connection
• After tensorflow.js model is loaded, no server connection is needed
• Powered by HTML5, WebRTC, Tensorflow.js

한국어 설명

Tested Environment

Chrome - 71.0.3578.98

How to use

1. Configuration

Please allow chrome to use "Camera".
If you want to auto download for recorded video, Please allow "Automatic Downloads".

2. Object detection

To test object detection, click "Test Object Detection" button. Loading TFJS model take 1~2 minutes.
This button draw bounding box to video.

3. Smart Recording

To start smart recording, click "Smart Recording" Button.
During smart recording mode,
If a object is detected, Recording is started.
If the object disapear for 3 seconds, Recording is stoped.
If recording is continued for 1 minute, record video is created and recording is re-started.

4. Recording

To start recording, click "Recording" Button.
This button start recording regardless of object detection. If recording is continued for 1 minute, record video is created and recording is re-started.

5. Navigator

5.1. video

Click "hamberger" button and check out the created video.
User can download or remove video. The video hold system memory until user remove it.
To save system memory, please download and remove video from navigator.

5.2. Auto download

If you plan to use long-term recording, Click auto "Auto Download & Remove" button.

5.3. download all and remove all

Download all and remove all videos