Fig. 5
From: Identification of plant-parasitic nematode genera in turfgrass using deep learning algorithms
Architectures of fine-tuned models used in this study. A EfficientNetV2-S containing a total of 10 FusedMBConv modules followed by 30 MBConv modules with a sample input image size of 3 × 384 × 384. B MobileNetV3-L model used in this study contains a total of 15 inverted residual bottlenecks with a sample input image of size 3 × 224 × 224. C ResNet101 model used in this study contains a total of 33 bottlenecks with a sample input image of size 3 × 224 × 224. D Swin V2-B model used in this study containing a total of 24 SwinTransformerBlock V2 modules and 3 PatchMergingV2 modules, with a sample input image of size 3 × 256 × 256.
