Shahab Nabavi
Shahab Nabavi
Hello, I have the same problem: Mean IoU over 150 classes: 0.0006 Pixel-wise Accuracy: 1.96%
Hello, @eddy-ilg Unknown bottom blob 'img0_a_org' (layer 'crop_params', bottom index 0) Is there any solution?
Hello @guanfuchen, Here is a link for detailed info of Res101-FCN. It is used in NIPS 2017 paper "Predicting Scene Parsing and Motion Dynamics in the Future" Link: https://papers.nips.cc/paper/7267-predicting-scene-parsing-and-motion-dynamics-in-the-future-supplemental.zip I...
Hello, Yes I have just realized that it doens't work. It only predicts class 0 and 2 with low mIoU.
@littlebelly @meetshah1995 I have made some changes in the code. Now the mean IoU is 75.7 which is very close to 79%. I will publish it very soon.
Hello, I can get 75.7. Do you want me to merge it with the current repo?
Hello, I have trained PSPNet with default configuration. I got 80% mean IoU. However, this mean IoU seems to be category IoU not class IoU!
I got these results from train_coarse_extra.py Yes this is my configuration
best record: [val loss 0.11416], [acc 0.97769], [acc_cls 0.86788], [mean_iu 0.80071], [fwavacc 0.95816], [epoch 36]
args = { 'train_batch_size': 8, 'lr': 1e-2 / (math.sqrt(16. / 8)), 'lr_decay': 0.9, 'max_iter': 10e4, 'input_size': 340, 'weight_decay': 1e-4, 'momentum': 0.9, 'snapshot': '', # empty string denotes learning from scratch...