Has anyone tried to do image recognition?

The place to discuss development topics.
Discussions on the RICOH THETA API and SDK, image processing of spherical images, other hardware related topics, introduction of useful libraries, etc.

Has anyone tried to do image recognition?

Postby SelWestern » 11 Aug 2017, 20:19

I'm looking into making an app that can recognize simple objects in an image. Just wondering if anyone else has tried this, or has had any experience with this using the Ricoh Theta camera and its SDK.
Posts: 1
Joined: 11 Aug 2017, 19:45

Re: Has anyone tried to do image recognition?

Postby jcasman » 14 Aug 2017, 15:53

The RICOH Cloud API may cover what you want. It is just being released, has a Python SDK, and includes several different types of image recognition: Human Detection - Detects human heads in an image; Face Detection - Detects human faces in an image; and Face Recognition - Compares faces in two images to enable developers to identify people. It's also certified with the highest accuracy level in the field. Main site: https://api.ricoh/ and main GitHub repo:

If you do try and do some work with it, any chance you could post your experiences on the Unofficial Guide? The THETA API Usage category would be a good spot to post: http://lists.theta360.guide/c/theta-api-usage Seriously, even just your general impressions would be useful. I'm wanting to evaluate it myself, so any comments welcome.
Posts: 193
Joined: 17 Apr 2015, 17:23

Re: Has anyone tried to do image recognition?

Postby CorLeone » 18 Aug 2017, 08:54

I tried the demo application uploading a few of my THETA images -- no face at all was recognized.
Lightroom had no problem identifying faces on these images.

In addition: Is this API capable to handle the equirectangular images of the THETA? Depending on the position in the image, faces might be heavily distorted or even split at the borders. The samples in the demo are just plain rectilinear.
Posts: 78
Joined: 20 Feb 2016, 01:18

Re: Has anyone tried to do image recognition?

Postby codetricity » 18 Aug 2017, 15:34

At a DeveloperWeek hackathon a few years ago, a team used the theta S and the Microsoft API to process facial recognition with emotions. I believe they did tests with both a dual-fisheye video stream and an equirectangular image. It was a long time ago and my memory is not so good.



I know that another community member is using a dual-fisheye video stream for object recognition processing.
Posts: 612
Joined: 31 Jul 2015, 01:56

Re: Has anyone tried to do image recognition?

Postby davids445 » 30 Aug 2018, 15:55

Google can recognize images without any app. You should try it.
tank trouble
Last edited by davids445 on 12 Jul 2019, 12:40, edited 1 time in total.
Posts: 2
Joined: 18 Jul 2018, 06:32

Re: Has anyone tried to do image recognition?

Postby JudyHart » 30 May 2019, 06:49

Definitely, there are many free apps and tools for that. Bespecular is one of the few I'm using for my design projects. If you need something universal (for text as well), KNFB Reader is the best variant. Look at the course materials: Multiplatform Mobile App Development with React Native.
Studying at The Hong Kong University of Science and Technology
Posts: 2
Joined: 31 Aug 2018, 07:56

Re: Has anyone tried to do image recognition?

Postby Swayer » 06 Jun 2019, 12:30

I do. It is really interesting. There are many free apps and tools for that. You know that computers can identify objects, people, places, and buy diamonds actions in images. Artificial Intelligence has a very good role in image recognition. It improves the quality of the image.
Posts: 7
Joined: 04 Jun 2019, 05:01

Return to Development

All times are UTC