Google Vision or Google AutoML for Identifying Matching Images based on Image Search

[Nirmal Kumar]

Dear Friends,

I would like to know which is the right product should i use for my scenario

1.I have 3000+ Images in Folder 1  ( some of the images are even cropped)

2.I would like to search a given Image (which is not a cropped one). Can GCV or GAML has any feature to identify the matching exact image from the 3000+ images folder and give me the matched filename ?

Note :

- Should i need to depend on Cloud Tools?

- Can this be accomplished using Tensor Flow or other Tools?

Regards

Nirmal

[Xinjie Zheng]

Hi Nirmal,

It's not yet available but we're woking on similarity search features that combine the best of both GCV and GAML.

It would be great if you would be able to share with us your use case/workflow (crop matching? searching within your product listing or other verticals? etc.) as that will help better shape our feature roadmap.

Thanks.

[Nirmal Kumar]

Hi Xinjie Zheng,

Here my detailed use case

1.I buy images/photographs from External Agencies which are licensed ones (for a year). These images are placed in a Folder 1. This has more than 3000 Images

2.My Content team, frequently takes some Images from this folder, Crops them and Publishes in our Multiple Websites.

3.After a year, some of the images published in the site may got expired. We want to know which are the images expired so that we can remove them from our site.

Solution :

1.Here the solution we thought of. Push all the 3000+ Original UNCROPPED Licensed images to GCV/Cloud Tools

2.Then pick each image from our website and search matching image in the GCV/Cloud Tools

Any idea how Google Cloud Tools can solve our business case?

Regards

Nirmal

[Ji Zhu]

Nirmal:

You may want to check the new product search api provided publicly: https://cloud.google.com/vision/product-search/docs/

On Friday, August 3, 2018 at 8:17:05 AM UTC-7, Nirmal Kumar wrote: