Table 7.
Research benchmarks with other medical object detection with CNNs techniques.
| Author | Method | Object | mAP |
|---|---|---|---|
| Rezvy et al. [17] | Modified Mask-RCNN | Endoscopy disease | 0.51 |
| Lin et al. [19] | CNNs-based YOLACT | Psoriasis skin | 0.85 |
| Shenavarmasouleh et al. [25] |
Mask-RCNN | Exudates and microaneurysms | 0.43 |
| Pai et al. [26] | VGG16 and Mask-RCNN |
Brain tumor | 0.90 |
| Masood et al. | Mask-RCNN | Brain tumor | 0.94 |
| Cai et al. [28] | Mask-RCNN | Pulmonary nodule | 0.88 |
| Chiao et al. [29] | Mask-RCNN | Breast tumor | 0.75 |
| Vuola et al. [30] | Mask-RCNN U-Net Ensemble |
Nuclei | 0.52 |
| Long et al. | Probability-Mask-RCNN | Pulmonary embolism | 0.81 |
| Our proposed model | ResNet50 and Mask-RCNN | Fetal heart defect | 0.98 |