We've significantly redesigned the recognition system, making it much more accurate. Now, description and keyword selection work together, significantly improving results.
Video Recognition
You can now recognize videos not only in IMS Studio, but also directly in the web version (mp4)! Recognition now takes multiple frames into account, resulting in more accurate results.
Hints
You can now provide hints during image processing for even more accurate results. For example, you can more specifically tell the neural network where the footage was filmed or what exactly you want to see in the description. To do this, click "Refine prompt" in the file selection panel. If you want alternative titles and descriptions for already recognized files, simply click "Refine and regenerate."
Models for every taste!
Our team believes everyone should have a choice... Choice of recognition model! We present 3 models to choose from:
- Base (1 recognition)
- Extra (2 recognitions)
- Thinking (5 recognitions). You can see what's on its mind by clicking on the light bulb.
The names speak for themselves. You can select a model in the upper right corner (next to the settings).
ℹ️Now, keyword selection and title/description generation with the base model require only 1 recognition!
Flexible settings
The settings for description and title generation have also changed. More options for what to do with the existing description have appeared. Now you can not only keep it or replace it, but also use it as a suggestion, or insert a generated description before or after it.
We've moved the description settings to the other settings in a single window. To edit them, click the "Settings" button in the upper right corner.
On this wonderful note, we conclude our overview of our smart update. Subscribe to us on social networks (https://t.me/imstocker_en - here we publish our news) and write your suggestions and recommendations in telegram chat (https://t.me/imstocker_chat_en) or send a feedback form (https://imstocker.com/en/page/feedback?product=ims-vision). See you soon, dear users)