Dataset

pedest.jpg

Pedestrian

pedest2.jpg

Pedestrian

Skater

Pedestrian

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

from clevrml import

Image_Model

import os

key = os.environ[ "API-KEY" ]

model = Image_Model( )

model.edit_model(

   api_key=key,

   class_names=

   example_folders=

   model_name=

)

<Pending Input>,

<Pending Input>,

"null"

Output

Your model has been updated!

Details:

-------------------------------------

Model Name: 

Deployed: True

Build: Complete

Cost: $0.99

------------------------------------

JSON: 

{

   Model_Name:

   Deployed: True,

   Build: "Complete",

   Cost: 0.99

}

"null"

folder_edited.png

/Home/Pedestrians/

Now that our data has been uploaded, let's run the code snippet to update our model.

Click the "Run" button to send a request to the Image Model API.

Let's add the new data of pedestrians to our model's "Memory". Click on all the images of a pedestrian.  

Hint: Click on all the images with a yellow border.

loading.gif

When building AI models, requirements change rapidly and updates need to be made. We need to make an update our model to recognize images of pedestrians.

 

With Neural Networks, this would require us collecting another 10,000 images of pedestrians and executing another training loop for hours or days. On clevrML, all we have to do is collect 3 more images of pedestrians and add them to our model's "Memory" (Hence, having an Active Memory ). Within 10 seconds of runtime, our model will be updated.

Our model has been updated successfully. Now if we show our model images of pedestrians, it will be able to recognize the images correctly. This type of dynamic editing is only possible with clevrML's Active Memory Learning technology. With Neural Networks, it is currently impossible to achieve this without building a completely new model.