TuSimple/mx-maskrcnn

Issues

• Step 3 fix

  #151 opened 6 years ago by umyta
  1

• Unable to download pretrain_model?

  #154 opened 6 years ago by Kobeee0614
  1

• error: ‘Fill’ was not declared in this scope

  #152 opened 6 years ago by zhufan123
  4

• what is the difference between ROIAlign and ROIAlign_v1 you provide?

  #153 opened 6 years ago by Danee-wawawa
  0

• find no difinition of Fill

  #116 opened 7 years ago by hengshan123
  2

• input data format

  #150 opened 6 years ago by lrj16
  0

• Train

  #149 opened 6 years ago by 994374821
  0

• ROIAlign bug

  #148 opened 6 years ago by eli2014
  0

• train_maskrcnn bug

  #147 opened 6 years ago by MichaelLv
  0

• ROIAlign Bug

  #146 opened 6 years ago by MichaelLv
  0

• mx-maskrcnn for keypoint detection

  #144 opened 6 years ago by tianyuecao
  1

• How to fit segmentation to keypoint detection?

  #131 opened 7 years ago by hdjsjyl
  1

• Dummy Data Inference

  #142 opened 6 years ago by xinyu-intel
  1

• AssertionError: P5_lateral_weight not initialized

  #145 opened 6 years ago by chowkamlee81
  0

• out_shape larger than data_shape

  #139 opened 6 years ago by tianyuecao
  1

• Speed drop when training?

  #143 opened 6 years ago by rxqy
  2

• The speed of train_maskrcnn on coco

  #118 opened 7 years ago by zl1994
  3

• How to add roiAlign to mxnet and no need to compile source code of mxnet?

  #121 opened 7 years ago by hdjsjyl
  1

• How to continue a training after sudden break，such as blackout？

  #141 opened 7 years ago by hahakid
  0

• RoI assign level bug

  #138 opened 7 years ago by YeWenting
  1

• ground truth roi load

  #140 opened 7 years ago by cloudfool
  0

• How to use it as Retinanet without instance segmentation?

  #106 opened 7 years ago by chinakook
  1

• nan!!!!!!

  #137 opened 7 years ago by dingfg
  0

• After changeing some of the params, the speed of convergence when training RPN is amazing

  #136 opened 7 years ago by zl1994
  0

• gtFine_labelTrainIds.png does not exist

  #135 opened 7 years ago by dingfg
  1

• mxnet compile error "fatal error: ../mshadow_op.h"

  #123 opened 7 years ago by 7oud
  5

• demo and demo_single_image

  #134 opened 7 years ago by cloudfool
  1

• why should fix parameters in training maskrcnn?

  #124 opened 7 years ago by gaosanyuan
  4

• Fatal error happened during compile. Missing ../mshadow_op.h

  #133 opened 7 years ago by 139419qw
  1

• Running error

  #130 opened 7 years ago by qilicun
  2

• How to select the area threshold which is used to assign level of proposal from RPN?

  #132 opened 7 years ago by hdjsjyl
  0

• Is it possible to train for keypoint detection?

  #120 opened 7 years ago by TalkIsCheapGiveMeMoney
  1

• the bbox of the large object is not as exact as the smaller object

  #125 opened 7 years ago by gaosanyuan
  3

• Can I deploy mx-maskrcnn in the ubuntu14.04？

  #100 opened 7 years ago by lt573152873
  1

• bounding box accuracy

  #129 opened 7 years ago by perp
  0

• why bn_data_moving_var is so large after training the model

  #127 opened 7 years ago by gaosanyuan
  4

• Is it reasonable to set the PIXEL_MEANS in rcnn/config.json file to (0,0,0)?

  #113 opened 7 years ago by MudeLin
  2

• Could you tell me how to make my own data??

  #128 opened 7 years ago by jstzsjw
  1

• On the rcnn_fc_bbox_weight_test and rcnn_fc_bbox_bias_test

  #126 opened 7 years ago by gaosanyuan
  1

• If I want to use it to train face alignment model, Do I need to change roidb? What's the meaning of 'max_classes' and 'max_overlaps'?

  #117 opened 7 years ago by hdjsjyl
  1

• On rpn_relu

  #119 opened 7 years ago by gaosanyuan
  0

• Will python 3 and windows be supported?

  #109 opened 7 years ago by mailqq100
  1

• End-to-end training?

  #115 opened 7 years ago by ysfalo
  0

• mask-rcnn how to allocate memory, If my own dataset has some big size images like 7127*7127, it has cudo memory error.

  #114 opened 7 years ago by hdjsjyl
  2

• unexpected keyword argument 'allow_extra'

  #104 opened 7 years ago by apchenstu
  1

• while I use my train model to test my dataset, error is " operands could not be broadcast together with shapes (0,0) (174,193)"

  #111 opened 7 years ago by napolun279
  1

• run demo-mask by gpus？

  #110 opened 7 years ago by ChangliangSun
  0

• about PIXEL_MEANS.

  #107 opened 7 years ago by franciszzj
  1

• The output of ROIPoolingAlign layer

  #102 opened 7 years ago by Mendel1
  2

• Inference on multiple images (batchSize>1)

  #101 opened 7 years ago by bubbles1990
  0