tensorflow-deeplab-resnet

tensorflow-deeplab-resnet copied to clipboard

Running evaluation on images of different size

1

Hello, I am trying to create a service that evaluates images. To improve performance I am trying to use the same network on all images. I use inference.py as a...

burhanbvk

[use the npy2ckpt.py to transfer my own resnet50 pre-train model]

1

I want to use the npy2ckpt.py to transfer my own resnet50 pre-train model: the layer name in my pre-train resnet50 model are: bn4c_branch2c bn5b_branch2b res3d_branch2b res2b_branch2b res2c_branch2a bn4b_branch2a res4b_branch2a bn4a_branch2c...

y-kl8

Resource exhausted: OOM when allocating tensor with shape[144,12,20,2048]

1

I only set batch_size=1,if batch_size>1 ,error occured: when batch_size=1,and other parameters is default, the loss does not down,below is loss step 19960 loss = 1.322, (0.337 sec/step) step 19961 loss...

coderclear

Evaluation errors submitting my test results to the PASCAL server

1

I test the pascal voc 2012 data, and submit my test results to the evaluation serber. I got the following Evaluation errors: http://host.robots.ox.ac.uk:8080/eval/view_submission/17144/ Anybody knows how to solve it? Thanks!

mokii

Failed precondition: VOCdevkit

6

2018-03-31 15:16:50.186012: W tensorflow/core/framework/op_kernel.cc:1158] Failed precondition: VOCdevkit 2018-03-31 15:16:50.186013: W tensorflow/core/framework/op_kernel.cc:1158] Failed precondition: VOCdevkit step 1 loss = 0.252, (1.038 sec/step) step 2 loss = 0.200, (0.209 sec/step) 2018-03-31 15:16:51.357591:...

jiyongma

High accuracy 0.843 in train_msc val?

1

Train for 20K iterations with original parameters in train_msc.py with about 10K VOC training set and val the model.ckpt-8000 for 1449 val images but got the high accuracy:0.843? The training...

mokii

Some other changes in the `train.py` have been made but they could rather be ignored. Although the part where I restore everything but the last layer is interesting if one...

mgarbade

fix inference.py for TF 1.0 && reformat using pep8 code style

5

- fix inference.py for TF 1.0 - reformat using pep8 code style

corenel