Loading request...
The user requests a model with a smaller input size (e.g., 512x512) because the current 1024x1024 input size uses too much memory. They suggest releasing models with smaller input sizes or converting the pretrained model to ONNX using a 512x512 input size.
The input size 1024x1024 uses too much memory for my case, would you consider release models with smaller input size, such as 512x512? Or directly convert your pretrained model (pytorch?) to onnx using 512x512 input size might also work well. With the static onnx model, it seems very hard to change the input size.