Image Quality Sorter¶
The Image Quality Sorter tool automatically assesses the Quality of each item and assigns a corresponding value to the Pick Label property in the database based on:
bandiera rossa per Rifiutata
bandiera gialla per In sospeso.
bandiera verde per Accettata
The Image Quality Sorter assigns Pick Labels to items according to the default settings in
, or the dedicated options set in this view.The quality score to determine the Pick Label property can be evaluated in one of two ways: using four basic factors for each image (Blur, Noise, Exposure, and Compression), or using a deep learning neural network engine.
The first approach helps to determine whether the basic factors have compromised image quality, however the output depends on fine-tuning made the user and it does not include any assessment of the aesthetic value of the image.
The second approach uses an artificial intelligence engine based on this model to predict an aesthetic score. As deep-learning is an end-to-end solution, it doesn’t require hyper-parameter settings, which makes this approach far easier to use.
Nota
Since quality assessment can be a time consuming process, it’s a good idea to check Work on all processor cores and restrict the job to certain albums or tags.
All the Image Quality Sorter settings are described in this section of the manual.
The Scan Mode settings configure how the database information for the items from the selected collection will be processed. Clean all and re-scan resets all data and scans all items from scratch, Scan non-assigned only will be faster to process just items that have not previously been assigned Pick Labels.
While the image quality sorting process is underway, a progress indicator is displayed in the bottom right corner of the main window.
I risultati dell’ordinamento qualità compariranno nella vista Etichette della barra laterale destra.
Importante
To run properly, the process needs access to the deep-learning model that can be downloaded at the first run of digiKam. See the Quick Start section of this manual for details.