facebookresearch/unbiased-teacher issues and pull requests

#83 - What are the versions of PyTorch and CUDA？

Issue - State: open - Opened by wmjlincy about 2 years ago

#82 - Why is the “out_features" in the fully connected layer num_classes+1 when calculates cls_score?

Issue - State: closed - Opened by wmjlincy about 2 years ago

#81 - How can I use pascal VOC as evalutation? Because the default is "COCOeval".

Issue - State: open - Opened by bliblioscar about 2 years ago

#80 - RuntimeError: CUDA error: the launch timed out and was terminated

Issue - State: closed - Opened by fedyhajali over 2 years ago - 1 comment

#79 - .

Issue - State: closed - Opened by MMarvasti over 2 years ago

#78 - About verification, the accuracy is 0

Issue - State: open - Opened by mary-0830 over 2 years ago - 3 comments

#77 - Hello, Could you help me with this proplem please?

Issue - State: open - Opened by EdwardDo69 over 2 years ago - 2 comments

#76 - Hello, can you help with this error please?

Issue - State: open - Opened by tysnow2022 over 2 years ago - 1 comment

#75 - Experiments without EMA and focal loss

Issue - State: open - Opened by Mansoor-at almost 3 years ago

#74 - Uploaded PASCAL VOC weight's cocoEVAL performance

Issue - State: closed - Opened by honggyuchoi almost 3 years ago - 1 comment

#73 - Splits percentage (Quick question - easy to answer)

Issue - State: closed - Opened by fabianfallasmoya almost 3 years ago - 2 comments

#72 - Supervised Loss for Student

Issue - State: closed - Opened by michaelku1 almost 3 years ago - 2 comments

#71 - Data splits in form of COCO image ids

Issue - State: open - Opened by vadimkantorov almost 3 years ago

#70 - Random crops enabled?

Issue - State: open - Opened by vadimkantorov almost 3 years ago - 2 comments

#69 - Question about pseudo-labels generation

Issue - State: open - Opened by vadimkantorov almost 3 years ago - 10 comments

#68 - Effect of PreciseBN

Issue - State: closed - Opened by vadimkantorov about 3 years ago - 2 comments

#67 - How to calculate the Pseudo-label accuracy?

Issue - State: closed - Opened by ASAKARUM about 3 years ago - 2 comments

#66 - Does anyone train successfully on LVIS dataset ?

Issue - State: open - Opened by seanzhuh about 3 years ago

#65 - Twice evaluation during BURN_UP_STEP

Issue - State: closed - Opened by qdd1234 about 3 years ago

#63 - ImportError: cannot import name 'FastRCNNOutputs'

Issue - State: closed - Opened by devillove084 about 3 years ago

#62 - TypeError: init() takes from 5 to 6 positional arguments but 7 were given

Issue - State: closed - Opened by devillove084 about 3 years ago

#61 - How to get the AP0.5 result on VOC dataset?

Issue - State: open - Opened by Complicateddd about 3 years ago

#60 - About some questions on training organized own dataset

Issue - State: open - Opened by dhhcj about 3 years ago - 1 comment

#59 - About organized own dataset

Issue - State: closed - Opened by dhhcj about 3 years ago - 1 comment

#58 - Random erasing fully erasing objects

Issue - State: open - Opened by guilhermecorreagv about 3 years ago

#57 - Issue of load pretrain model

Issue - State: open - Opened by winnerziqi about 3 years ago

#56 - Loss rise problem

Issue - State: open - Opened by Jackson-zhao97 over 3 years ago - 2 comments

#55 - How to run on multi-node?

Issue - State: open - Opened by machengcheng2016 over 3 years ago

#54 - A little question about bs

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

#53 - lower batch size ： 2 labeled + 2 unlabeled -----> bad AP

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

#52 - gt_instances is NoneType object

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

#51 - Any guide for registering and using custom dataset?

Issue - State: open - Opened by sarmientoj24 over 3 years ago - 2 comments

#50 - teacher model does not in eval mode?

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

#49 - Details about the traning

Issue - State: open - Opened by MendelXu over 3 years ago - 6 comments

#48 - About the training iterations of supervised model and semipervised model

Issue - State: open - Opened by Maggie-000 over 3 years ago

#47 - "batch norm layers are not updated as well" for teacher

Issue - State: closed - Opened by Divadi over 3 years ago - 2 comments

#46 - loss Nan

Issue - State: open - Opened by Fly-dream12 over 3 years ago - 4 comments

#45 - Fix for: It was assumed that image_id would be in range(len(dataset))

Pull Request - State: open - Opened by JoshAusHessen over 3 years ago - 2 comments
Labels: CLA Signed

#44 - ImportError: cannot import name 'FastRCNNOutputs'

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

#43 - How to evaluate the performance of labeled data?

Issue - State: closed - Opened by wjn922 over 3 years ago - 1 comment

#42 - Implementation of CSD

Issue - State: closed - Opened by wjn922 over 3 years ago - 1 comment

#41 - Question about burn in steps

Issue - State: closed - Opened by wjn922 over 3 years ago - 1 comment

#40 - COCOEval on Pascal dataset experiments

Issue - State: closed - Opened by JongMokKim over 3 years ago - 1 comment

#39 - How to calculate the Box Accuracy in Fig 4a?

Issue - State: closed - Opened by GeoffreyChen777 over 3 years ago - 2 comments

#38 - How to load weights from Detectron2 model zoo?

Issue - State: closed - Opened by jjbuck over 3 years ago - 2 comments

#37 - Loss_box_reg starts at zero and increases during training

Issue - State: closed - Opened by ogamache over 3 years ago - 2 comments

#36 - Performance reproduction : learning rate step

Issue - State: open - Opened by JongMokKim over 3 years ago - 7 comments
Labels: Discussion

#35 - Change two stages to one stage

Issue - State: open - Opened by Jackson-zhao97 over 3 years ago - 7 comments

#34 - Visualize the pseudo-labels

Issue - State: closed - Opened by ogamache over 3 years ago - 2 comments

#33 - Question about using pseudo labels supervision

Issue - State: closed - Opened by zehuichen123 over 3 years ago - 1 comment

#32 - Easiest way to use this for supervised object detection as baseline?

Issue - State: closed - Opened by sarmientoj24 over 3 years ago - 3 comments

#31 - Issue About EMA process in code

Issue - State: closed - Opened by Yuuuuuuuuuuuuuuuuuummy over 3 years ago - 1 comment

#30 - Augmentation?

Issue - State: closed - Opened by jiangzhengkai over 3 years ago - 2 comments

#29 - How can I infer?

Issue - State: open - Opened by yeonsikch over 3 years ago - 7 comments
Labels: enhancement

#28 - I want to make sure I'm right about what I know

Issue - State: closed - Opened by yeonsikch over 3 years ago - 4 comments

#27 - update student model

Issue - State: closed - Opened by hachreak over 3 years ago - 2 comments

#26 - About VOC dataset training settings.

Issue - State: closed - Opened by xpyqiubai over 3 years ago - 1 comment

#25 - question: double evaluation during training

Issue - State: closed - Opened by hachreak over 3 years ago - 1 comment

#24 - TypeError: 'NoneType' object is not iterable

Issue - State: closed - Opened by xiaohui0225 over 3 years ago - 1 comment

#23 - self.iter restart from zero when I resume a training

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

#22 - About COCO_supervision.txt

Issue - State: closed - Opened by jiangzhengkai over 3 years ago - 1 comment

#21 - how to get labeled data in coco-standard

Issue - State: closed - Opened by namihagi over 3 years ago - 1 comment

#20 - How to perform NMS before generating pseudo labels?

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

#19 - How to inference on my onw pictures?

Issue - State: open - Opened by sisrfeng over 3 years ago - 4 comments
Labels: enhancement

#18 - Data loader implementation issue

Issue - State: closed - Opened by vlfom over 3 years ago - 1 comment

#17 - same idea found in cvpr2021

Issue - State: closed - Opened by AndyYuan96 over 3 years ago - 7 comments

#16 - Confusion about the loss of second stage

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

#15 - Master

Pull Request - State: closed - Opened by ZhuoranYu almost 4 years ago - 1 comment

#14 - using randon select date get more higher result than using COCO_supervision.txt

Issue - State: closed - Opened by happyxuwork almost 4 years ago - 3 comments

#13 - Run in custom dataset

Issue - State: closed - Opened by wangerxiao001 almost 4 years ago - 2 comments

#12 - a question about the code: `TEST.VAL_LOSS`

Issue - State: closed - Opened by xiaohu2015 almost 4 years ago - 2 comments

#11 - Share the evaluated json files

Issue - State: closed - Opened by aashusri almost 4 years ago - 1 comment

#10 - Partially labelled dataset

Issue - State: open - Opened by Fabio-Arup-Panella almost 4 years ago - 3 comments

#9 - TypeError: _get_local_path() got an unexpected keyword argument 'force'

Issue - State: open - Opened by zhangnan-hust almost 4 years ago

#8 - Experiment on custom dataset

Issue - State: closed - Opened by A1exy almost 4 years ago - 1 comment

#7 - Nan appears when training

Issue - State: closed - Opened by happyxuwork almost 4 years ago - 3 comments

#6 - Question about burn-in steps

Issue - State: closed - Opened by Chrisfsj2051 almost 4 years ago - 2 comments

GitHub / facebookresearch/unbiased-teacher issues and pull requests