yoxu515/aot-benchmark

Issues

• Recreating the general examples

  #67 opened 2 months ago by johnathanchiu
  2

• metric

  #83 opened 3 months ago by yaoyao57
  0

• Optimizing Model Performance: Exploring ONNX Export and Engine Integration with TensorRT and OpenVino

  #68 opened a year ago by AntonioConsiglio
  4

• Does this model convert to onnx to run on edge device?

  #82 opened 3 months ago by SuyueLiu
  1

• To feed truth values into the model during the testing phase?

  #81 opened 4 months ago by lbbcalm
  4

• How to reproduce the pre_dav_ytb result?

  #80 opened 6 months ago by pengqinher
  0

• INT_MAX `local2global`

  #79 opened 7 months ago by bhack
  0

• Static image

  #78 opened 8 months ago by xujinzh
  0

• How to train aot? Where is aot's train.py?

  #77 opened 8 months ago by 22236
  0

• SDPA Attention

  #76 opened 9 months ago by bhack
  0

• F.interpolate upsample gradient with amp

  #75 opened 9 months ago by bhack
  1

• SAM encoder/decoder

  #39 opened 9 months ago by bhack
  2

• Work Around for Pytorch-Correlation-extension

  #69 opened a year ago by ziqipang
  5

• 0.5px/1px E/SE misaligment

  #47 opened 2 years ago by bhack
  10

• Would it help if I had more "ground truth" masks?

  #65 opened a year ago by rihns38
  2

• Evaluator Memory

  #74 opened a year ago by bhack
  0

• AUX loss WEIGHT and RATIO

  #72 opened a year ago by bhack
  3

• cfg.DIST_ENABLE false fail

  #36 opened 2 years ago by bhack
  21

• [PAOT] swinv2b

  #73 opened a year ago by bhack
  0

• about DMAOT details?

  #62 opened a year ago by zhanghongyong123456
  8

• package specified for utils

  #71 opened a year ago by bingnoi
  1

• lamda parameter which indicates the neighborhood size of 11 by 11

  #61 opened a year ago by chiru83
  1

• Demo.py with more then one GPU

  #70 opened a year ago by Renaud38
  2

• VOTSTS dataset

  #55 opened a year ago by iremonur
  2

• How to use the MS-AOT to test custom dataset

  #42 opened 2 years ago by xwhkkk
  5

• Object merging in tracking

  #66 opened a year ago by leonid-pishchulin
  3

• Mask size compared to image size affects tracking

  #46 opened 2 years ago by rihns38
  5

• YoutubeVOS quantitative numbers

  #64 opened a year ago by Amshaker
  0

• [Reproducing quantitative numbers of AOT and DeAOT]

  #63 opened a year ago by Amshaker
  7

• I get an error when I run tools/demo.py

  #53 opened a year ago by Dagoli
  2

• MS-Deaot

  #60 opened a year ago by bhack
  2

• About tracking.

  #59 opened a year ago by nilesh23041999
  0

• Gated Propagation Module - ID Branch and LT/ST propagation

  #58 opened a year ago by AntonioConsiglio
  3

• Question about short update memory function while using DeAOTT

  #57 opened a year ago by AntonioConsiglio
  5

• How to train DeAOT?

  #56 opened a year ago by yongliu20
  8

• Hello, I would like to ask some questions about AOT architecture

  #52 opened a year ago by zhanghongyong123456
  1

• The modules that work on tiny objects

  #50 opened 2 years ago by xwhkkk
  1

• Dynamically remove items from tracking

  #54 opened a year ago by Davidyao99
  1

• 请问如何在不训练阶段1的情况下，直接训练阶段二，是否可以用你们提供的静态模型（AOTT_PRE.pth）作为预训练模型？

  #49 opened a year ago by zhanghongyong123456
  2

• Shape mismatch when training

  #48 opened 2 years ago by martinarroyo
  4

• How to run only main training without pre-training

  #43 opened 2 years ago by fiskrt
  1

• CUDA out of memory

  #44 opened 2 years ago by xwhkkk
  2

• Cannot train model youtubeVOS. Bug in attention.py?

  #45 opened 2 years ago by fiskrt
  3

• Tensorboard log on VAL

  #41 opened 2 years ago by bhack
  0

• No module named spatial_correlation_Sampler

  #40 opened 2 years ago by sivaji123256
  2

• Disantangle long term memory update and reference frame

  #37 opened 2 years ago by bhack
  18

• How to increase the efficiency of propagation modules?

  #38 opened 2 years ago by MartinSmeyer
  2

• F score over the boundaries

  #34 opened 2 years ago by bhack
  4

• AOT/DeAOT perfs on MOSE dataset

  #33 opened 2 years ago by bhack
  5

• Evaluate with more sparse supervision

  #35 opened 2 years ago by bhack
  4