SLaK

SLaK copied to clipboard

Incorrect number of parameters

2

Dear Authors, In the README.md the number of parameters mentioned for SLaK-T with 51x51 kernels is ~30M however, when I ran your eval code for Imagenet with your trained model,...

shashankskagnihotri

您好，您在论文中有做Object detection的实验，并在项目中提到可以用mmdetection框架做目标检测，请问可以更新下detection相关的代码？

920666358

About mmseg

1

I can not find depthwise_conv2d_implicit_gemm.py. Py file not found_ depthwise_ conv2d_ implicit_ gemm_ C( import _depthwise_conv2d_implicit_gemm_C as _extension), depthwise_conv2d_implicit_gemm was not found in the slak.py (from depthwise_conv2d_implicit_gemm import DepthWiseConv2dImplicitGEMM), please...

tthupup

Hi there! This project is very cool! I see you currently share the model weights with DropBox. Would you be interested in sharing your models in the [Hugging Face Hub](http://hf.co/models)?...

osanseviero

Dear Authors, In the Object Segmentation and Dectection tab of your README.md you mention in step 4: Download and use our weights. For examples, to evaluate SLaK-tiny + UperNet on...

shashankskagnihotri