facebookresearch/adaptive_teacher

Issues

• FloatingPointError: Predicted boxes or scores contain Inf/NaN

  #57 opened a year ago by anranbixin
  9

• How to conduct ablation experiments, set the corresponding Loss to 0, or remove the corresponding module from the model?

  #63 opened 2 years ago by Donghuan777
  2

• FloatingPointError: Predicted boxes or scores contain Inf/NaN. Training has diverged.

  #71 opened a year ago by ouyang11111
  2

• gtfine problem

  #72 opened a year ago by PlutoNov
  0

• TypeError: register_buffer() takes 3 positional arguments but 4 were given

  #55 opened 2 years ago by HaroldHuanrongLIU
  3

• How to train a SourceOnly model or Oracle model?

  #62 opened a year ago by Donghuan777
  2

• About VGG16 pre-trained on ImageNet

  #36 opened 2 years ago by pengjw23
  3

• FileNotFoundError: [Errno 2] No such file or directory: 'dataseed/COCO_supervision.txt'

  #70 opened a year ago by Manjuphoenix
  0

• RuntimeError: CUDA error: no kernel image is available for execution on the device

  #65 opened 2 years ago by Manjuphoenix
  1

• The shorter side of the image is resized to 800 when testing？

  #69 opened a year ago by SesshomaruL
  0

• Selecting custom classes from dataset

  #68 opened a year ago by Manjuphoenix
  0

• Viewing tensor board running instance on docker and some doubts

  #67 opened a year ago by Manjuphoenix
  0

• demo

  #66 opened 2 years ago by wansd
  1

• Save teacher weight

  #64 opened 2 years ago by ShenZheng2000
  1

• How to get the AP of each label?

  #59 opened 2 years ago by judyxt
  2

• How to visualize the predicted results on pictures using the pretrained model?

  #61 opened 2 years ago by Donghuan777
  0

• How to train with the coco format datasets?

  #54 opened 2 years ago by Donghuan777
  1

• How

  #60 opened 2 years ago by Donghuan777
  0

• backbone modification

  #58 opened 2 years ago by achintski
  1

• The Beta parameter of Foggy Cityscapes

  #56 opened 2 years ago by lch1999
  1

• Error while loading pretrained model weights from `detectron2://ImageNetPretrained/MSRA/R-101.pkl` for training with custom dataset

  #49 opened 2 years ago by hnanacc
  0

• question about selecting best model (validation or post last step?)

  #51 opened 2 years ago by shjustinbaek
  1

• what is crop ratio?

  #52 opened 2 years ago by victor00070
  1

• AttributeError: 'NoneType' object has no attribute 'keys'

  #30 opened 2 years ago by zwang123
  6

• How to open the output file

  #53 opened 2 years ago by fatherplus
  1

• resnet backbone dont converge

  #48 opened 2 years ago by victor00070
  1

• how to change backbone to resnet

  #47 opened 2 years ago by victor00070
  1

• Can adaptive teacher use one-stage detector?

  #34 opened 2 years ago by huanmx
  3

• Teacher or Student model at inference time?

  #39 opened 2 years ago by cmckee13
  1

• Why 2 seperate annotation files?

  #45 opened 2 years ago by victor00070
  4

• Very low performance during evaluation

  #46 opened 2 years ago by victor00070
  0

• Can not reproduce the results of "cityscapes" to "foggy cityscapes".

  #24 opened 3 years ago by cwhgn
  12

• How to calculate mAP of model and mAP of each class, I tried to use sample code, but can not, how to solve?

  #43 opened 2 years ago by zhezhejunyyds
  1

• Memory Leak

  #44 opened 2 years ago by GeoffreyChen777
  1

• why were the mAP and AP50 getting smaller with the increase of iterations

  #41 opened 2 years ago by sysuzgg
  1

• mAP predicted by student_model is sometimes higher than teacher

  #42 opened 2 years ago by firekeepers
  7

• About discriminator in the paper

  #38 opened 2 years ago by YangJae96
  1

• detectron2 v0.6 has removed FastRCNNOutputs class

  #40 opened 2 years ago by alvinti
  1

• Can not reproduce the results on "foggy cityscapes" due to Out of Memory Issue

  #33 opened 2 years ago by onkarkris
  7

• possible to wrap the teacher model by DistributedDataParallel?

  #35 opened 2 years ago by Weijiang-Xiong
  2

• Why your test scale is larger than 600?

  #37 opened 2 years ago by Yorionice1
  3

• some puzzles about using "branch.startswith("supervised")" in adapteacher/modeling/meta_arch/rcnn.py

  #32 opened 2 years ago by gedebachichi
  3

• The AP and APl is 0.000 when evaluation in cityscapes_foggy_val

  #31 opened 2 years ago by Doris1231
  1

• The checkpoint state_dict contains keys that are not used by the model

  #28 opened 2 years ago by ad45675
  3

• Pre-trained VGG16 on ImageNet not shared?

  #29 opened 2 years ago by mecarill
  1

• AP NaN

  #27 opened 2 years ago by darkhan-s
  4

• Why your DA-FRCNN implementation uses multi-scale training trick?

  #23 opened 3 years ago by tmp12316
  3

• Optimal number of learning iterations

  #26 opened 2 years ago by miurakenichi
  10

• Loss becomes NaN

  #25 opened 2 years ago by miurakenichi
  3

• Evaluation for each category

  #22 opened 3 years ago by helq2612
  2