Finder Mode versus Single Shot Mode: How to manage the search modes in CraftAR

icon Date FAQ

icon Date Dec 15, 2020

icon author Catchoom Team

icon folder , , , , ,

CraftAR offers two types of search modes in order to pass queries to the SDKs and search the image database:

Both search modes are supported in all our Image Recognition SDKs, plugins, and libraries. Deciding which scan mode you would like to implement in your mobile or web app depends on your use case and user experience goals.

In this article, we describe the main differences between the two search modes and provide some pointers to useful tutorials.

Single Shot Mode search

Single Shot mode performs a search with a single image. The device won’t start scanning automatically when the camera is open. Therefore, it leaves more time for the user to prepare the device and point at the specific object. To perform the scan, the user would need to point at the object and tap the screen or click a button to scan.

By calling this method, the SDK triggers the still image capture from the camera and forwards the captured image to the component in charge of performing the image recognition search (the Cloud Image Recognition module or the On-device Image Recognition module).

Implementing with Cloud Image Recognition

Each time the scanning is initiated, one query is sent to our CraftAR Cloud Service. As a Cloud Visual Scan is a request sent from your app to the CraftAR Cloud Service to recognize an object.

Calculating with an approximate 10-15 KB size for each query, it translates to around 10-15 KB data consumption for your user for each scanning event.

Useful tutorials to read:

Implementing with On-Device Image Recognition

When using Single Shot mode with On-Device Image Recognition, the queries are managed locally.

Useful tutorials to read:

Finder Mode search

CraftAR allows searching continuously for matches while the user just points (or tosses) the camera around – we call it Finder Mode. After opening the camera, Finder Mode starts to search continuously with no need for user intervention, at a controllable rate.

You can set a time limit for the search, after which the search stops if there was no successful recognition up until that point. Then, you can alert or instruct your user. When used on Cloud collections, this helps limit the amount of visual scans consumed without a successful result.

Implementing with Cloud Image Recognition

As a rule of thumb, using the Finder Mode in Cloud Image Recognition with the default setting of up to 2 queries sent per second. This leads to a maximum 30 KB of data consumed per second (calculated with 2 queries of 10-15 KB sent per second) for your end user as an estimate.

However, you may control the rate of continuous visual search (control the time between one query sent to another, with a maximum of 2 seconds in between).

As a Cloud Visual Scan is a request sent from your app to the CraftAR Cloud Service to recognize an object, regardless the result, all the queries sent within one session are counted towards your Cloud Visual Scan quota.

Useful tutorials to read:

Implementing with On-device Image Recognition

When using Finder Mode with On-Device Image Recognition, the queries are managed locally, so no data is consumed on your user’s side for sending the queries.

Useful tutorials to read:

Related Posts in FAQ