RED-7669: optimize OCR-module performance

* reset test and settings
This commit is contained in:
Kilian Schuettler 2023-11-15 16:00:00 +01:00
parent 4c225c2219
commit 2632d2023d
2 changed files with 4 additions and 4 deletions

View File

@ -12,9 +12,9 @@ import lombok.experimental.FieldDefaults;
@FieldDefaults(level = AccessLevel.PRIVATE)
public class OcrServiceSettings {
int ocrThreadCount = 16; // Number of OCR threads
int ocrThreadCount = 4; // Number of OCR threads
int imageExtractThreadCount = 2; // Number of image extraction threads
int gsProcessCount = 5; // Number of Ghostscript processes
int gsProcessCount = 2; // Number of Ghostscript processes
int dpi = 300; // Target DPI for binarized images
int psmOverride = -1; // Overrides the page segmentation mode if > 0
int minImageHeight = 20; // Minimum height for images to be processed

View File

@ -31,7 +31,7 @@ import io.micrometer.prometheus.PrometheusMeterRegistry;
import io.micrometer.prometheus.PrometheusTimer;
import lombok.SneakyThrows;
//@Disabled // Ghostscript/Tesseract/Leptonica is not available on build server. If you want to run the test install these dependencies. See README.md for help.
@Disabled // Ghostscript/Tesseract/Leptonica is not available on build server. If you want to run the test install these dependencies. See README.md for help.
@SpringBootTest()
public class OcrServiceIntegrationTest extends AbstractTest {
@ -64,7 +64,7 @@ public class OcrServiceIntegrationTest extends AbstractTest {
@SneakyThrows
public void testOcr() {
String text = testOCR("files/10.SYN524464 FS (A16148C) - Absorção cutânea.pdf");
String text = testOCR("files/2009-1048395_50pages_tables.pdf");
}