GeorgeSeif/Semantic-Segmentation-Suite issues and pull requests

#248 - Why do you use an image cropping function instead of resizing?

Issue - State: closed - Opened by ghost over 3 years ago

#247 - Updated to newest version of tf_slim

Pull Request - State: closed - Opened by ghost over 3 years ago

#246 - ModuleNotFoundError: No module named 'tensorflow.contrib'

Issue - State: open - Opened by hailuu684 over 3 years ago

#245 - Plotting Error While Doing Continue Training

Issue - State: open - Opened by hmza09 over 3 years ago - 1 comment

#244 - Requirements for this project?

Issue - State: open - Opened by suresh-pokharel over 3 years ago - 2 comments

#243 - docs: fix simple typo, trainble -> trainable

Pull Request - State: closed - Opened by timgates42 over 3 years ago

#242 - Pre-trained Checkpoints

Issue - State: open - Opened by NaeemKhan333 almost 4 years ago - 1 comment

#241 - is crop size same with image size for production ?

Issue - State: open - Opened by ramdhan1989 about 4 years ago - 1 comment

#240 - Tensorflow v2

Pull Request - State: closed - Opened by jschaffer-edr about 4 years ago

#239 - CONTINUE_TRAINING

Issue - State: open - Opened by lyc1995452-star over 4 years ago

#238 - Batch size changes output with same images

Issue - State: open - Opened by RorryB over 4 years ago

#237 - Librarier functions

Issue - State: open - Opened by Olliver2020 over 4 years ago - 1 comment

#236 - Forecast picture is all black

Issue - State: open - Opened by xumeng19 over 4 years ago - 1 comment

#235 - Bibtex format for cite this project

Issue - State: open - Opened by jnakanojp over 4 years ago

#234 - "_gt" images in checkpoint folder look different than they actually are

Issue - State: open - Opened by ersinsiginc over 4 years ago - 1 comment

#233 - Hello~I want ask one question.Why use the crop function to adjust images to input network rather than use the resize function?

Issue - State: open - Opened by Shiny-ZhangXiXin over 4 years ago - 1 comment

#232 - How to apply the weight loss to this project?

Issue - State: open - Opened by Classmate-Huang over 4 years ago

#231 - Is it only applicable for GPU ?

Issue - State: closed - Opened by ramdhan1989 over 4 years ago - 1 comment

#230 - Error for Crop Height and Width of the FC-DenseNet103?

Issue - State: open - Opened by AI-ML-Enthusiast over 4 years ago

#229 - [Q] Why is there "resize_bilinear" for "image_level_features"? (I think it does nothing..)

Issue - State: closed - Opened by ywpkwon over 4 years ago - 1 comment

#228 - Prediction takes too long

Issue - State: open - Opened by Arunavameister over 4 years ago - 1 comment

#227 - question about frontend models

Issue - State: open - Opened by ghost over 4 years ago - 3 comments

#226 - failed to download the pretrained-weights in get_pretrained_checkpoints.py

Issue - State: open - Opened by l99500 almost 5 years ago - 2 comments

#225 - cant load checkpoint files

Issue - State: open - Opened by skywo1f almost 5 years ago - 10 comments

#224 - AdapNet implementation

Issue - State: open - Opened by kame-hameha almost 5 years ago

#223 - What is the output modified using one-hot code if I want to train model by 3D images? Such as video frames

Issue - State: open - Opened by che006 almost 5 years ago

#222 - run test.py and predict.py

Issue - State: open - Opened by INJNainggolan about 5 years ago

#221 - there is wrong when transfer ckpt to pb model

Issue - State: closed - Opened by cena001plus about 5 years ago - 1 comment

#220 - how

Issue - State: closed - Opened by cena001plus about 5 years ago

#219 - How to train it well?

Issue - State: open - Opened by mengxia1994 about 5 years ago - 1 comment

#218 - bug about MultiResolutionFusion in Refinenet.py

Issue - State: closed - Opened by cena001plus about 5 years ago - 1 comment

#217 - Using matrix instead of image

Issue - State: open - Opened by pavva94 about 5 years ago

#216 - how to use gpu?

Issue - State: open - Opened by mengxia1994 about 5 years ago - 3 comments

#215 - working with ccp dataset

Pull Request - State: closed - Opened by victordeleau over 5 years ago

#214 - how to run test.py?

Issue - State: open - Opened by vkes2541 over 5 years ago - 3 comments

#213 - Can a master change the prediction code to the camera real-time prediction code?

Issue - State: open - Opened by wanghengmm over 5 years ago

#212 - Is there only a code that predicts a single sheet, is there a code that the camera splits in real time?

Issue - State: open - Opened by wanghengmm over 5 years ago

#211 - Restoring Checkpoint for Predict

Issue - State: open - Opened by preetkhaturia over 5 years ago - 1 comment

#210 - question about class balance

Issue - State: open - Opened by YyChen818 over 5 years ago - 1 comment

#209 - why i increase the bathsize, miou decrease?

Issue - State: open - Opened by memeda2232 over 5 years ago

#208 - Loss spikes and per-class accuracy low on roads dataset

Issue - State: open - Opened by jakefleisig over 5 years ago

#207 - Minor syntax issues fixed

Pull Request - State: open - Opened by kant over 5 years ago

#206 - Why depricated? Are u tired George?

Issue - State: open - Opened by sunjinr over 5 years ago

#205 - when I ran train.py.I have the problem:subprocess.CalledProcessError: Command '['python', 'utils/get_pretrained_checkpoints.py', '--model=ResNet101']' returned non-zero exit status 1

Issue - State: open - Opened by sweetboxwwy over 5 years ago - 22 comments

#204 - error: the following arguments are required: --image, --checkpoint_path, --model

Issue - State: closed - Opened by Watebear over 5 years ago

#203 - --epoch_start_i must be 0

Issue - State: open - Opened by HangHuang over 5 years ago - 2 comments

#202 - gt pictures in checkpoints different with val_labels?

Issue - State: closed - Opened by x7hkvip over 5 years ago - 3 comments

#201 - Pretrained weights not available

Issue - State: closed - Opened by jesarte over 5 years ago - 5 comments

#200 - Conv2d/Relu order in Refinenet

Issue - State: open - Opened by Attila94 over 5 years ago

#199 - This repo doesn't work anymore (I tested 4 models but the results were not as mentioned in the description)

Issue - State: closed - Opened by developpeur3d over 5 years ago

#198 - providing your own checkpoints in drive

Issue - State: open - Opened by yacineDZ44 over 5 years ago

#197 - I'm interested in it. Can anyone tell me the steps to run it?

Issue - State: open - Opened by qingHX over 5 years ago - 1 comment

#196 - Cannot load any model

Issue - State: open - Opened by alex88o over 5 years ago - 8 comments

#195 - Not working for input crop size other than 512 X 512

Issue - State: open - Opened by suddu22 over 5 years ago - 4 comments

#194 - How to split batch between multiple GPUs for prediction.

Issue - State: open - Opened by ttocs167 over 5 years ago - 3 comments

#193 - Resnet18 and Xception39 for BiSeNeT

Issue - State: open - Opened by RakshithGB over 5 years ago - 2 comments

#192 - how to train on cityscapedataset ? it always pops up error of " line 45, in load_image image = cv2.cvtColor(cv2.imread(path,-1), cv2.COLOR_GRAY2RGB) #BGR2RGB cv2.error: OpenCV(3.4.3) /io/opencv/modules/imgproc/src/color.cpp:181: error: (-215:Assertion failed) !_src.empty() in function 'cvtColor' " seeking help

Issue - State: closed - Opened by nooriahmed over 5 years ago - 1 comment

#191 - Maybe I do something wrong？current loss = nan

Issue - State: open - Opened by tianwen0110 over 5 years ago - 2 comments

#190 - training setting infomation

Issue - State: open - Opened by merrybingo over 5 years ago

#189 - Using GPU?

Issue - State: closed - Opened by jsolves over 5 years ago - 2 comments

#188 - Removing OpenCV Dependency?

Issue - State: closed - Opened by fzyzcjy over 5 years ago - 2 comments

#187 - Issue running with mut1ny dataset

Issue - State: closed - Opened by ecidon over 5 years ago - 1 comment

#186 - PSPNet Image Sizes

Issue - State: open - Opened by RorryB over 5 years ago

#185 - Prediction Directly

Issue - State: open - Opened by RihamHazem over 5 years ago - 1 comment

#184 - How to create and save the actual TF SavedModel? [Feature requests]

Issue - State: open - Opened by ukiras123 over 5 years ago

#183 - How to use VGG as front end in deeplabv3+ ?

Issue - State: open - Opened by nikhilparmar over 5 years ago

#182 - Error during plotting when continuing training

Issue - State: open - Opened by TomasRiker over 5 years ago - 1 comment

#181 - Potential Batch Norm Architecture Bug for Encoder-Decoder

Issue - State: open - Opened by k22jung over 5 years ago - 2 comments

#180 - how do I use input data with shape (width, height, 2) or 4 channels?

Issue - State: open - Opened by apple1986 over 5 years ago

#179 - How to get the RGB image of each predicted segment?

Issue - State: open - Opened by MoAbd over 5 years ago - 3 comments

#178 - how to achieve real-time semantic segmentation of video

Issue - State: open - Opened by sry984789463 over 5 years ago

#177 - Removing background and using larger image

Issue - State: open - Opened by shubhank008 over 5 years ago - 1 comment

#176 - Error importing the image - !_src.empty()

Issue - State: closed - Opened by freddifederica over 5 years ago - 2 comments

#175 - current losss is always 0.00000

Issue - State: open - Opened by RobinHan24 over 5 years ago - 11 comments

#174 - current_loss is always 0.0000

Issue - State: open - Opened by RobinHan24 over 5 years ago - 5 comments

#172 - Ignoring classes in the class_dict.csv

Issue - State: closed - Opened by k22jung over 5 years ago - 2 comments

#171 - ResourceExhaustedError when using batch_size >= 64

Issue - State: open - Opened by vanqm over 5 years ago - 3 comments

#168 - Training small custom dataset

Issue - State: open - Opened by jsan3386 almost 6 years ago - 8 comments

#166 - Adding tensorflow summary metrics support for Tensorboard

Pull Request - State: open - Opened by dpsm almost 6 years ago - 2 comments

#165 - about the input image size

Issue - State: open - Opened by l99500 almost 6 years ago - 3 comments

#162 - About RefineNet detail

Issue - State: closed - Opened by Spritea almost 6 years ago - 2 comments

#161 - What does the training set do

Issue - State: closed - Opened by junxiaoge almost 6 years ago - 1 comment

#158 - DDSC : could not convert BatchDescriptor {count: 1 feature_map_count: 1024 spatial: 0 0 value_min: 0.000000 value_max: 0.000000 layout: BatchDepthYX} to cudnn tensor descriptor: CUDNN_STATUS_BAD_PARAM Aborted

Issue - State: open - Opened by TaoZhong11 almost 6 years ago - 1 comment

#156 - Unable to run training on BiSeNet using pre-trained MobileNetV2

Issue - State: open - Opened by anzisheng almost 6 years ago - 10 comments

#153 - i am chinese student. so i can not download the pre- train wights and the checkpoints .i need them to Tune in code . can you send if to me? this is my email [email protected] ,thank you very much!!

Issue - State: open - Opened by wjsw1781 almost 6 years ago - 1 comment

#152 - DDSC Value Error using ResNet101

Issue - State: open - Opened by MaxKrapp almost 6 years ago - 5 comments

#148 - Fail to load model checkpoint weights when using other frontends than ResNet101.

Issue - State: open - Opened by akhaloo almost 6 years ago - 7 comments

#147 - Multi-gpu training support

Issue - State: open - Opened by vikjet almost 6 years ago - 3 comments

#145 - The question in the Model BiSeNet.

Issue - State: open - Opened by Reagan1311 almost 6 years ago - 7 comments

#120 - the problem about BiSeNet

Issue - State: open - Opened by csufangyu almost 6 years ago - 13 comments

#118 - Error saving plots after first epoch (x or y dimension incorrect)

Issue - State: closed - Opened by roelofvandijk almost 6 years ago - 2 comments

#114 - Predict.py Line54

Issue - State: closed - Opened by Darknesszlx almost 6 years ago - 2 comments

#111 - About label_image

Issue - State: open - Opened by Leo551 almost 6 years ago - 5 comments

#105 - Trained model into a pb file

Issue - State: open - Opened by gulingfengze about 6 years ago - 5 comments

#100 - update: A few improvements for `train.py`

Pull Request - State: closed - Opened by mrshu about 6 years ago - 1 comment

#99 - update: Add crop to predict.py

Pull Request - State: closed - Opened by mrshu about 6 years ago

#98 - I changed model input logits shape as (64，128，128，1) ， labels shape as (64，128，128，1) for binary classify ( I do not want to use one hot label )，which loss function will work？ I used tf.nn.sigmoid_crossentropy_with_logits() but did not work，refer to tensorflow document which explain logits has same shape to labels ，I am wrong? and your lovazs loss not work for refinenet resnet50 binary classify( I didnt try other model)

Issue - State: closed - Opened by liuchuanloong about 6 years ago - 7 comments

#97 - predict.py example

Issue - State: closed - Opened by SerinaWei about 6 years ago - 2 comments

#96 - something wrong when generate an image

Issue - State: closed - Opened by 1453042287 about 6 years ago - 2 comments

#95 - test.py

Issue - State: closed - Opened by kazu82018303 about 6 years ago

GitHub / GeorgeSeif/Semantic-Segmentation-Suite issues and pull requests