shinya7y/UniverseNet

Issues

• Welcome update to OpenMMLab 2.0

  #39 opened 2 years ago by vansin
  0

• How to save the predicted bbox in a file?

  #37 opened 2 years ago by nile649
  1

• How to calculate mAP for each category for waymo 2D object detection

  #36 opened 2 years ago by gopikrishnabs
  7

• YOLOv4 of mmdet. version

  #35 opened 3 years ago by Tim-Hung
  1

• f'{obj_type} is not in the {registry.name} registry') KeyError: 'WaymoOpenDataset is not in the dataset registry' and python -m torch.distributed.launch --nproc_per_node=$GPUS --master_port=$PORT \ SyntaxError: invalid syntax

  #34 opened 3 years ago by gopikrishnabs
  11

• How can i use yolox with

  #33 opened 3 years ago by betterhalfwzm
  1

• Pre-trained checkpoints on Waymo Open Dataset

  #32 opened 3 years ago by KaiChen1998
  1

• Thread for minor results

  #18 opened 4 years ago by shinya7y
  3

• visualization technologies

  #31 opened 3 years ago by sure7018
  1

• Extract outputs from various layers of model

  #30 opened 3 years ago by JosephAssaker
  1

• UniverseNet Pretrained on Manga109

  #29 opened 3 years ago by JosephAssaker
  1

• Test picture

  #28 opened 3 years ago by sure7018
  4

• Train other datasets

  #27 opened 3 years ago by sure7018
  6

• inference annotations as strings?

  #26 opened 3 years ago by rainwala
  1

• UniverseNet using experience

  #22 opened 3 years ago by HBioquant
  1

• Hardware requirements for trainning Waymo Open Dataset

  #25 opened 3 years ago by anhhna
  3

• Multi gpu training

  #24 opened 3 years ago by sh7jacobi
  1

• loss nan error

  #13 opened 3 years ago by whut2962575697
  8

• I try to fine-tuning from a COCO pre-trained model，but the loss is always nan.

  #12 opened 3 years ago by muzishen
  8

• There are many memory occupation in GPU0

  #23 opened 3 years ago by John-Yao
  1

• Have you tried with the latest resnet model?

  #16 opened 4 years ago by Manojbhat09
  5

• the training duration of relationnet ++

  #21 opened 4 years ago by sure7018
  9

• Is there an object relation module in your code

  #20 opened 4 years ago by sure7018
  1

• Have you tried using autoassign in RelationNet++?

  #19 opened 4 years ago by zhongqiu1245
  2

• Have you tried PAA assign as anchor assigner?

  #17 opened 4 years ago by HAOCHENYE
  1

• Any technical reports?

  #2 opened 4 years ago by lucasjinreal
  2

• loss nan error when set filter_empty_gt=False

  #14 opened 4 years ago by zhengye1995
  5

• Group error

  #15 opened 4 years ago by CuongNN218
  1

• Pretrained model for universenet in Nightowls dataset

  #11 opened 4 years ago by rsj007
  9

• Very low AP when using cfg.data.val to build the dataset

  #10 opened 4 years ago by Jossome
  5

• Request for NightOwls Model weights

  #9 opened 4 years ago by shardulparab97
  3

• AssertionError: Default process group is not initialized

  #8 opened 4 years ago by iampartho
  2

• RuntimeError: The size of tensor a (4) must match the size of tensor b (7) at non-singleton dimension 1

  #7 opened 4 years ago by Ericargus
  4

• when use model inference, RuntimeError: input image is smaller than kernel.How can I solve the error?

  #3 opened 4 years ago by chenzyhust
  7

• For universenet is it possible to use pretrained weights while changing the class number from 80 to 1?

  #4 opened 4 years ago by iampartho
  6

• time and data_time

  #6 opened 4 years ago by swapnil-saha
  1

• Weirdly low AP at early epochs of fine-tuning. Could you please share your log file ?

  #5 opened 4 years ago by HongQWang
  9

• COCO weights for UniverseNet+GFL

  #1 opened 4 years ago by amirassov
  2