How to create an endpoint using the model trained by Vertex AI?

[Batukush]

I trained a model using Google Cloud Vertex AI. I have a dataset containing different videos and model does the video classification of these videos. In the Google Cloud Platform, I can find accuracy of the given videos after model has been trained. However, I could not find a way to create an Endpoint which is required to implement my model to the project.

In other words, I have a model trained by Google Cloud’s Vertex AI and I want to create an Endpoint which accepts a request with video and returns the accuracy according to the trained model. Google Cloud provides a way to do it using Endpoints but as shown below, following error occurs. This model cannot be deployed to an Endpoint

I could not find any documents on why this error is given to me. On the other hand, when I download the given model, code requires input type as below with the video;

“RESNETISH_V3: float32 Tensor, shape=(None, 128)”.

I do not know what is this variable type and this is basically where I am stuck. I need help deploying my trained model to any serverless environment, create a code handling posted video and return the accuracy and class of the given video to the request.

[anjelab]

As this would be hard to reproduce on my end without access on your project, I suggest to raise a 1:1 support and a dedicated resource with access on your project can help you with your issue.

[rashearth]

Hi, I also have the same issue when working with time prediction. If you have any solution, please kindly let us know!

[rashearth]

Just for supplementary information, here is whats happening . Please take a look, i cant click on the deploy buttton.

 

[athletech]

hey I have the same issue. was anyone able to resolve this deployment problem? 

[athletech]

Ended up finding the answer in documentation.  https://cloud.google.com/vertex-ai/docs/video-data/classification/get-predictions