tianzhi0549/fcos issues and pull requests

#91 - TypeError: init() got an unexpected keyword argument 'confidence_threshold'

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

#90 - how low it takes to train the model

Issue - State: closed - Opened by NiNiZhang17 over 5 years ago - 2 comments

#89 - centerness shared head

Issue - State: open - Opened by AllenPeng0209 over 5 years ago - 10 comments

#88 - about the model

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

#87 - A QUESTION about input size

Issue - State: closed - Opened by qianzhang2018 over 5 years ago - 7 comments

#86 - comparsion with yolov3?

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

#85 - Have you train and test on VOC?

Issue - State: open - Opened by BlueBlueFF over 5 years ago - 27 comments

#84 - 能问下怎么网络可视化？

Issue - State: closed - Opened by qianzhang2018 over 5 years ago - 2 comments

#83 - samll object

Issue - State: closed - Opened by qianzhang2018 over 5 years ago

#82 - questions about the result

Issue - State: open - Opened by qianzhang2018 over 5 years ago - 7 comments

#81 - Should I adjust area_ranges to the input size?

Issue - State: closed - Opened by BlueBlueFF over 5 years ago - 4 comments

#80 - Is anyone has ever tried other classification loss?

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

#79 - IndexError: list index out of range

Issue - State: closed - Opened by qianzhang2018 over 5 years ago - 6 comments

#78 - How to do inference by multiple GPUs？

Issue - State: closed - Opened by ZhengMengbin over 5 years ago - 6 comments

#77 - About using syncbn.

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

#76 - how to set multiple regression levels for smaller image size?

Issue - State: closed - Opened by liwei46 over 5 years ago - 2 comments

#75 - Using EfficientNet as a backbone

Issue - State: open - Opened by lukasschmit over 5 years ago - 6 comments

#74 - Does anyone replace FL with GHM-C in FCOS ？It seems to be very low

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

#73 - About ground truth centerness

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

#72 - A question about IOUloss

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

#71 - error in install step, python setup.py

Issue - State: closed - Opened by Kangzf1996 over 5 years ago - 13 comments

#70 - out of memory

Issue - State: closed - Opened by heiyuxiaokai over 5 years ago - 8 comments

#69 - error about /FCOS/maskrcnn_benchmark/_C.cpython-36m-x86_64-linux-gnu.so

Issue - State: closed - Opened by ZhengyuZhang96 over 5 years ago - 2 comments

#68 - Create test_pr

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

#67 - Why use exp function in regression branch

Issue - State: open - Opened by lih627 over 5 years ago - 7 comments

#66 - Problem about centerness for trarning and inference

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

#65 - Problem about size mismatch.

Issue - State: closed - Opened by jsjs0827 over 5 years ago - 2 comments

#64 - Compile error using MSVC

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

#63 - when I fit the code to my dataset, all the losses are nan other than first 20 iterations

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

#62 - Compile error

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

#61 - coco path

Issue - State: closed - Opened by mama110 over 5 years ago - 3 comments

#60 - recall is relatively low

Issue - State: closed - Opened by Annihilation7 over 5 years ago - 1 comment

#59 - CalledProcessError: Command '['which', 'g++']' returned non-zero exit status 1.

Issue - State: closed - Opened by mengxzh15 over 5 years ago - 1 comment

#58 - How to measure total training memory?

Issue - State: closed - Opened by stigma0617 over 5 years ago - 1 comment

#57 - Convert the model to tensorflow

Issue - State: closed - Opened by TTMRonald over 5 years ago - 29 comments

#56 - 600ms on mx150 which is slow

Issue - State: closed - Opened by lucasjinreal over 5 years ago - 10 comments

#55 - about the learning rate

Issue - State: closed - Opened by jewelc92 over 5 years ago - 1 comment

#54 - How to train my own datasets (format is like coco datasets)

Issue - State: closed - Opened by Xavier-Zeng over 5 years ago - 51 comments

#53 - Is the range of target size related to the stride for a specific level

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

#52 - where is inference_single_cv.py file ?

Issue - State: closed - Opened by 3AceShowHand over 5 years ago - 2 comments

#51 - _C.cpython-37m-x86_64-linux-gnu.so

Issue - State: closed - Opened by fengfengfeng96 over 5 years ago - 2 comments

#50 - AttributeError: 'list' object has no attribute 'resize'

Issue - State: closed - Opened by jewelc92 over 5 years ago - 8 comments

#49 - RuntimeError: cannot perform reduction function min on tensor with no elements because the operation does not have an identity

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

#48 - About the dockerfile

Issue - State: closed - Opened by jewelc92 over 5 years ago - 1 comment

#47 - RuntimeError?

Issue - State: closed - Opened by ZhengMengbin over 5 years ago - 9 comments

#46 - 【Two situations】When compiling FCOS: (1) collect2: fatal error: cannot find 'ld'. (2) unable to execute 'x86_64-conda_cos6-linux-gnu-gcc': No such file or directory.

Issue - State: closed - Opened by Haonan-Qin over 5 years ago - 8 comments

#45 - what is the final loss?

Issue - State: closed - Opened by 3AceShowHand over 5 years ago - 2 comments

#44 - About the head feature sharing

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

#43 - ValueError: num_samples should be a positive integeral value, but got num_samples=0

Issue - State: closed - Opened by qianzhang2018 over 5 years ago - 4 comments

#42 - runtimeError: cannot perform reduction function min on tensor with no elements because the operation does not have an identity

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

#41 - Can you provide dockerfile?

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

#40 - training error about GC？

Issue - State: closed - Opened by ZhengyuZhang96 over 5 years ago - 6 comments

#39 - How to resume training？

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

#38 - /FCOS/maskrcnn_benchmark/layers

Issue - State: closed - Opened by qianzhang2018 over 5 years ago - 1 comment

#37 - /usr/local/cuda-9.0/include/crt/host_config.h:119:2: error: #error -- unsupported GNU version! gcc versions later than 6 are not supported! #error -- unsupported GNU version! gcc versions later than 6 are not supported!

Issue - State: closed - Opened by qianzhang2018 over 5 years ago - 1 comment

#36 - box target?

Issue - State: closed - Opened by YongtaoGe over 5 years ago - 15 comments

#35 - Can not download the pre-trained model

Issue - State: closed - Opened by ZhengMengbin over 5 years ago - 11 comments

#34 - training error, cannot start

Issue - State: closed - Opened by song6299 over 5 years ago - 4 comments

#33 - What is scale in class FCOSHead in focs.py?

Issue - State: closed - Opened by lg920810 over 5 years ago - 2 comments

#32 - a problem about ground-truth center-ness generation for inference

Issue - State: closed - Opened by gaoyuchris over 5 years ago - 10 comments

#31 - Improvement for small object detection

Issue - State: open - Opened by shadyatscu over 5 years ago - 12 comments

#30 - training doesn't cost time

Issue - State: closed - Opened by sjtu-YizhouYao over 5 years ago - 4 comments

#29 - compile error with pytorch1.0.0 nightly

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

#28 - When I train resnet101 backbone,I encounter this error,while I can train resnet50 successfuly with the same coco file,what should I do?

Issue - State: closed - Opened by gittigxuy over 5 years ago - 7 comments

#27 - How to reduce the cpu ocupation?

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

#26 - when i use the test_net.py to eval the model, an error happen

Issue - State: closed - Opened by houweidong over 5 years ago - 9 comments

#25 - How to handle the negative values in bbox prediction convolution layer when calculating the IOU Loss?

Issue - State: closed - Opened by Hwang64 over 5 years ago - 1 comment

#24 - when test image, the score of some object is low.

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

#23 - some question about loss.py in fcos file

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

#22 - centerness loss is larger than other loss

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

#21 - something wrong when install

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

#20 - Could FCOS overfit one single train image

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

#19 - 您好，请问为何在inference_single_cv中没有使用centerness

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

#18 - Question on the num_class inconsistent between fco head and fco postprocessor

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

#17 - Performance improvement from 36.6 to 37.1

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

#16 - What's the inference speed?

Issue - State: closed - Opened by auroua over 5 years ago - 7 comments

#15 - training error

Issue - State: closed - Opened by Baby47 over 5 years ago - 10 comments

#14 - Data preparation error

Issue - State: closed - Opened by Hwang64 over 5 years ago - 4 comments

#13 - about bceloss

Issue - State: closed - Opened by GGjerry01 over 5 years ago - 6 comments

#12 - wrong memory number in README

Issue - State: closed - Opened by zimenglan-sysu-512 over 5 years ago - 1 comment

#11 - How many time i need to train a converged model?

Issue - State: closed - Opened by houweidong over 5 years ago - 6 comments

GitHub / tianzhi0549/fcos issues and pull requests