PSPNet-tensorflow

PSPNet-tensorflow copied to clipboard

I have trained my PSP net model using this repo: https://github.com/holyseven/PSPNet-TF-Reproduce . I wanted to load it through your inference code. I trained my for 3 classes on ADE_20K dataset....

lakshmankanakala

How to better do batch inference on multiple images

4

Hi there! My apologies if this is a stupid question as I am still very new to TF. I am trying to use PSP50 (pre-trained on ADE20K) for inference on...

jin-zhe

inference result with model trained with train.py is not good and model size is almost twise bigger then pretrained model

3

kiitosu

small size

1

Can I use the pre-training model to calculate the miou and mean Pixel Accuracy of a small-sized image, such as 128*256

mxkai

Mean image

24

May be the mean image in your code is fault. The mean image is Pascal.

FengLoveBella

Having trouble training

2

Hi, I'm trying to train on cityscapes in order to first replicate the 70% miou and then move to other driving datasets to see what happens. However, I'm having trouble...

AmeetR

When the network reaches convergence, The loss has been around 2.0

14

python train.py --is-training --update-mean-var --train-beta-gamma to run the network, but when the network reaches convergence, The loss has been around 2.0. How did you get 0.2? @hellochick Thank you very...

Hukongtao

training from scratch

1

Hi hellochick, i love your model. i would like to ask how many steps did you train in order to get to the model added in the repository? i would...

KodamaXYZ

evaluation mIoU: 0.00404 on cityscapes val dataset

7

Hi, @hellochick , I had run the evaluation command for cityscapes val dataset using your provided [trained model](https://drive.google.com/drive/folders/1JSrLS6f5OcRIm_WBQ_jD_VxLP3oOMSsH). However, the mIoU is too low (0.00404) which seems something wrong in...

amiltonwong

How to get (class) label IDs or train IDs for evaluation with cityscapes-scripts

1

If anyone has found a way to do this, could you please give me a hint? I guess there's simpler way than writing script to convert RGB segmentation output into...

ga84