search

GeorgeSeif / Semantic-Segmentation-Suite

Semantic Segmentation Suite in TensorFlow. Implement, train, and test new Semantic Segmentation models easily!
2.51k stars 878 forks source link

question about frontend models #227

Open ghost opened 4 years ago

ghost commented 4 years ago

im new to this field . In this repo, there are frontend models. if i use them, am i doing fine-tuning ? or transfer learning?
for example, if i use ResNet101 as frontend and u-net as model, am i doing transfer learning from ResNet101 to U-net ? Sorry to ask such a basic question.

mojoee commented 4 years ago

The frontend is just to extract the features from the image. The model itself will give the prediction by feeding in those extracted features

ramdhan1989 commented 4 years ago

@mojoee can we use others such as VGG or EfficientNet from keras application as the front end ?

mojoee commented 4 years ago

well that's fine, but makes it complicated because you need to make sure to transform your tensor from keras to the tensorflow format. I would say, it's better to stay with tensorflow.