Problems while running train_tracker.sh

[SMZCC]

When I run the file of train_tracker.sh,I got a problem :

[libprotobuf ERROR google/protobuf/text_format.cc:298] Error parsing text-format  caffe.SolverParameter: 12:14: Expected integer, got: RANDOM_SEED
WARNING: Logging before InitGoogleLogging() is written to STDERR
F1009 21:19:19.955904  3587 upgrade_proto.cpp:1101] Check failed:  ReadProtoFromTextFile(param_file, param) Failed to parse SolverParameter file: ./nets/solver.prototxt
 *** Check failure stack trace: ***
train_tracker.sh: 行 16:  3587 已放弃               (核心已转储) python -m goturn.train.train --imagenet $IMAGENET_FOLDER --alov $ALOV_FOLDER --init_caffemodel $INIT_CAFFEMODEL --train_prototxt $TRACKER_PROTO --solver_prototxt $SOLVER_PROTO --lamda_shift 5 --lamda_scale 15 --min_scale -0.4 --max_scale 0.4 --gpu_id 0

Does that mean the RANDOM_SEED not having a value?Why doesn't it have a value? In my expectation, RANDOM_SEED should have a value, what is wrong ?And how can I fix it ?

Thanks

[nrupatunga]

In the solver.prototxt remove this line random_seed: RANDOM_SEED

Note: Not tested from my end. Please do check.

[SMZCC]

• Can I remove that line without any effection to the whole program?
• What's the function of that line ? Or it is just a redundancy？ @nrupatunga

[nrupatunga]

Random seed is set here. So you can remove the line.

[SMZCC]

Thanks I have got that @nrupatunga

[SMZCC]

Dear nrupatunga, in the file of solver.prototxt , I find the parameter: net which confuses me in some way:

• the parameter, net, can't be removed, I have tried, then how to set it's value while we already have a init_model?
• Why did we need a init_model?
• Can't we train a model from scratch? Then how to do that?

  Thanks @nrupatunga

[nrupatunga]

I will clean up solver.prototxt, which will help you understand better. Thanks

[SMZCC]

• Thank you very much
• However, I'm now very curious about last three questions .I am eager for a key :sob: especially,how can I train a model without a init_model from scratch @nrupatunga

[nrupatunga]

@SMZCC The issue with solver.prototxt is fixed. Please do check the updated code.

the parameter, net, can't be removed, I have tried, then how to set it's value while we already have a init_model?

• This issue is fixed in updated code

Why did we need a init_model?

• This is out of scope of this discussion. For further understanding please refer here

Can't we train a model from scratch? Then how to do that?

• You can. Removing this line should work

[SMZCC]

Thank you very much @nrupatunga . I've got that.

You @nrupatunga are so strong that my way is still long.

[nrupatunga]

Thank you for testing. Closing this issue. Feel free to reopen if you find an issue