Introduction to nilvana™ Deep Detector

In this article we are going to cover the following topics:

No-Code AI Model! Let's see how to use it.

Please retrieve the IP address of the edge and make sure the edge is connected to the camera.

Click on the drop-down menu. To create a new project by clicking on the New Project button.

The operations will divide into 4 parts:

  1. Capture
  2. Annotate
  3. Train
  4. Inference

Capture Images From The Camera

Start from Capture, the frame of the camera shows in the middle of the page.

Click on the capture button to capture the current frame, and the frame will lists in the Images.

To preview the capture image by clicking on the name of image.

You can also delete the image through the trash can button next to image.

Switch the camera through the button above the frame of camera.

Also, close the camera through the button below the frame of camera.

Annotate Your Image

Start to annotate the capture images by switching to the Annotate page. Choose the images you want to annotate from the left list of Images.

To check the annotated images by switching the above filter to annotated, or keep annotating the unannotated images by the filter of unannotated.

Click and drag on the image to annotated the object, and insert the class name of the text area near the bounding box. The classes will show in the label list on the right after inserting, the new bounding box can directly use the inserted class name by clicking.

We also provide the annotation tools for you, you can:

  1. Fast switching the images.
  2. Zoom in or out of the image.
  3. Choose all bounding box of specific image
  4. Delete specific bounding box of an image

The annotation tools mentioned can also accomplish by the keyboard shortcut.

Don't forget to save the annotations.

Model Training

The next is Training, download the images and annotations.

We provide 2 different training ways for you, take the left option if you already have Nilvana workstation, you can also take the right option to train on Google Colab.

After finishing training, upload model to Deep detector and switch to Inference page.

Congratulations! Ready For Inference

In order to recognize different models, you can rename the model.

Choose the model you want to make inference in the left list. Please be patient, loading model need some time.

The inference result will show in the middle of the page.

At last, you can switch to different languages through the setting button below.

Contact us if you have any questions and suggestions through below question mark button.

Still need help? Contact Us Contact Us