Real-Time Scene Text Detection with Differentiable Binarization and Adaptive Scale Fusion

Open paper
Benchmark Model Rank Results
scene-text-detection-on-icdar-2015DBNet++ (ResNet-50) (1152)#14F-Measure: 87.3Precision: 90.9Recall: 83.9FPS: 10
scene-text-detection-on-icdar-2015DBNet++ (ResNet-18) (736)#28F-Measure: 83.1Precision: 90.1Recall: 77.2FPS: 44
scene-text-detection-on-msra-td500DBNet++ (ResNet-50) (736)#3F-Measure: 87.2Precision: 91.5Recall: 83.3FPS: 29
scene-text-detection-on-msra-td500DBNet++ (ResNet-18) (736)#5F-Measure: 85.1Precision: 87.9Recall: 82.5FPS: 55
scene-text-detection-on-msra-td500DBNet++ (ResNet-18) (512)#11F-Measure: 82.6Precision: 89.7Recall: 76.5FPS: 80
scene-text-detection-on-total-textDBNet++ (ResNet-50) (800)#9F-Measure: 86%Precision: 88.9Recall: 83.2FPS: 28
scene-text-detection-on-total-textDBNet++ (ResNet-18) (800)#16F-Measure: 83.3%Precision: 87.4Recall: 79.6FPS: 48

Rank counts only results with a code link.