Tim Meinhardt comments

Results 144 comments of


                                            Tim Meinhardt

I try to train with MOT17,when I use "multi_frame_attention = True",I get " RuntimeError: stack expects each tensor to be equal size, but got [2, 2, 21, 27, 43] at entry 0 and [2, 2, 21, 27, 42] at entry 1"

This will not work. Please follow the instructions closely. Just removing some of the options will of course change quite a lot. The CPU/GPU error indicates that our code is...

I can not reproduce the results

Cause you changed the batch size to 1. :) A different batch size means you must find new optimal learning rates and training epochs.

I can not reproduce the results

I do not know. You have to find new optimal LRs and epochs. A recommend starting point could be to half the learning rates just as you did with the...

I can not reproduce the results

Only the learning rates not the multiplicators (`lr_linear_proj_mult `). The weight decay can remain as it is. What appears different with different batch sizes? You mean why do you have...

I can not reproduce the results

Explaining the relation between batch size and learning rate goes beyond the support of this repository. :) You might have to adjust the number of epochs. But again you will...

ImportError: /home/simon/anaconda3/envs/trackFormer/lib/python3.7/site-packages/MultiScaleDeformableAttention-1.0-py3.7-linux-x86_64.egg/MultiScaleDeformableAttention.cpython-37m-x86_64-linux-gnu.so: undefined symbol: cudaSetupArgument

Our README only supports virtualenv and not conda. There is most likely a version mismatch. Did the installation of the MultiScaleDeformableAttention run through without errors?

ImportError: /home/simon/anaconda3/envs/trackFormer/lib/python3.7/site-packages/MultiScaleDeformableAttention-1.0-py3.7-linux-x86_64.egg/MultiScaleDeformableAttention.cpython-37m-x86_64-linux-gnu.so: undefined symbol: cudaSetupArgument

I have never encountered that error. So its hard to say. But my best guess is sth related to wrong package versions.

error in ms_deformable_im2col_cuda

This means you are running the model not on the same GPU architecture as you used for the compilcation of the `MultiScaleDeformableAttention` package. Please check your setup!

Table 4 and Table 5 using COCO pretraining or not?

I have asked the same question in a different issue. This line https://github.com/wjf5203/VNext/blob/d41c4a1b35e894df0ea90cce76f5e828d9a29da1/projects/IDOL/configs/ovis_r50.yaml#L3 seems to suggest that they used a model pretrained on COCO sequences. But I would appreciate a...

trained model？

Hello, the trained model files can be downloaded via ``` wget https://vision.in.tum.de/webshare/u/meinhard/e-osvos_models.zip unzip e-osvos_models.zip ``` Please follow the installation instructions in the README.md. Inference can be done on a single...