OCR that is ten times faster
The engine includes OCR that is ten times faster than Banco v1.
Introducing
Banco 1.5
Our most accurate pseudonymisation engine yet. Banco 1.5 replaces Banco v1. Available to everyone on 13 July.
Banco 1.5 and Banco 1.5 Pro are our new embedded local models: a major upgrade from Banco v1. They run on your computer without sending data, identify entity boundaries more accurately and produce more reliable mapping tables, faster.
| Metric | Banco v1 previous | Banco 1.5 ideal for everyday use | Banco 1.5 Pro the most complex documents | |
|---|---|---|---|---|
| Entities found | 90,3 % | 95,6 % | 99,0 % | Proportion of entities requiring anonymisation actually found by the engine |
| Overall precision | 86 % | 93,6 % | 98,9 % | Proportion of detections that are correct |
| Manual interventions per 100 pages | 12 | 3 | 2 | Number of corrections required by hand |
| Total time per 100 pages | 1,640 s | 430 s | 332 s | Engine time plus review time |
| Engine time per 100 pages | 80 s | 40 s | 72 s | Automatic processing time only |
| Documents processed per hour | 89 | 249 | 138 | Engine throughput, excluding review |
Banco v1 is Lexanon's original model. Banco 1.5 and Banco 1.5 Pro are both available in the application.
introducing banco 1.5 and 1.5 pro
Intelligent splitting of coordinated names
Banco 1.5 was developed to substantially improve organisation detection.
Faster analysis
Banco 1.5 also speeds up processing: up to 21 entities detected per second and 130 occurrences per second.
Better classification
We improved Banco 1.5's ability to distinguish categories. This improves grouping and reduces review time.
Better for everyday work
Banco 1.5 also receives a major usability update: PDF export is 95% faster, multimodal processing handles all file types, quality is improved and the engine is more resilient.
A broader range of documents
The new engine was developed using a wider range of documents than its predecessor, making it more resilient.