Input shape for pretrained model (models.detection.keypointrcnn_resnet50_fpn)

vision
1

import torch
from torchvision import models
from torchsummary import summary

model = models.detection.keypointrcnn_resnet50_fpn(pretrained=True)
model .eval()
summary(model , (3, 224, 224))

throws
AttributeError: ‘ImageList’ object has no attribute ‘size’

Question is what is the valid input_size for the summary api
Intent is to print the summary of the pre_trained_model