SpineNet-Pytorch

SpineNet-Pytorch copied to clipboard

zhanghaobucunzai

Training on single GPU

5

Is it possible to train RetinaNet with SpineNet-190 backbone on single GPU using the classic mmdetection command ```python tools/train.py ${CONFIG_FILE}```? I tried and I get this error: `Traceback (most recent...

Alexandre-Delplanque

`--------------------------------------------------------------------------- TypeError Traceback (most recent call last) in 9 # Build the detector 10 model = build_detector( ---> 11 cfg.model, train_cfg=cfg.train_cfg, test_cfg=cfg.test_cfg) 12 # Add an attribute for visualization convenience...

vishaldhull09

I want to use spinenet to train imagenet，and the input shape is（224， 224， 3），Then I met the error like this

Hukongtao

behaviour different in train and test

3

when train the model size is correct, while inference got error: ``` /spinenet.py", line 157, in forward target_feat = parent0_feat + parent1_feat RuntimeError: The size of tensor a (14) must...

lucasjinreal

From paper it claims that it supressed encoder-decoder pattern detection model, so that it can do FPN things in a single backbone, the experiements still using FPN?

lucasjinreal

sharing trained weights

1

hi @lucifer443 ,due to lack of enough resources i have used pretrained models to finetune(transfer learning) for [global wheat detection task](http://www.global-wheat.com/), and it would be very useful if i had...

farshid-mi

Hi @lucifer443 , Thanks for the pytorch implementation. Is there a Pytorch model of Spinenet pretrained on Imagenet (classification)? If I have to pre-train this classifier network using imagenet, what...

shruum