one stage train problem

[yangyongguang]

CUDA_VISIBLE_DEVICES=2 python ./pytorch/train.py train --config_path=./configs/car.fhd.onestage.config --model_dir=./save_model_one_stage_mask/ --resume=True Traceback (most recent call last): File "./pytorch/train.py", line 561, in fire.Fire() File "/home/sugon/.conda/envs/pytorch_env/lib/python3.6/site-packages/fire/core.py", line 127, in Fire component_trace = _Fire(component, args, context, name) File "/home/sugon/.conda/envs/pytorch_env/lib/python3.6/site-packages/fire/core.py", line 366, in _Fire component, remaining_args) File "/home/sugon/.conda/envs/pytorch_env/lib/python3.6/site-packages/fire/core.py", line 542, in _CallCallable result = fn(*varargs, **kwargs) File "./pytorch/train.py", line 168, in train text_format.Merge(proto_str, config) File "/home/sugon/.conda/envs/pytorch_env/lib/python3.6/site-packages/google/protobuf/text_format.py", line 574, in Merge descriptor_pool=descriptor_pool) File "/home/sugon/.conda/envs/pytorch_env/lib/python3.6/site-packages/google/protobuf/text_format.py", line 631, in MergeLines return parser.MergeLines(lines, message) File "/home/sugon/.conda/envs/pytorch_env/lib/python3.6/site-packages/google/protobuf/text_format.py", line 654, in MergeLines self._ParseOrMerge(lines, message) File "/home/sugon/.conda/envs/pytorch_env/lib/python3.6/site-packages/google/protobuf/text_format.py", line 676, in _ParseOrMerge self._MergeField(tokenizer, message) File "/home/sugon/.conda/envs/pytorch_env/lib/python3.6/site-packages/google/protobuf/text_format.py", line 801, in _MergeField merger(tokenizer, message, field) File "/home/sugon/.conda/envs/pytorch_env/lib/python3.6/site-packages/google/protobuf/text_format.py", line 875, in _MergeMessageField self._MergeField(tokenizer, sub_message) File "/home/sugon/.conda/envs/pytorch_env/lib/python3.6/site-packages/google/protobuf/text_format.py", line 768, in _MergeField (message_descriptor.full_name, name)) google.protobuf.text_format.ParseError: 108:3 : Message type "second.protos.InputReader" has no field named "max_number_of_voxels".

the other configs has all passed except this so could you give me some advice or offer me a new onestage.config.

[yangyongguang]

use car.fhd.config wo have get a 90.12%(easy) in val dateset,so ,I wanna to see the result by using one stage config,thank you.

[traveller59]

I forget to update car.fhd.onestage.config. actually the only difference between car.fhd.config and car.fhd.onestage.config is "rpn" in config file. it's very simple.

[yangyongguang]

so just change this line module_class_name: "RPNV2" to module_class_name: "RPN"?if not,can you send me a file. thank you for answering our questions on the go.

[traveller59]

I have update that file. you can compare it with car.fhd.config.

[yangyongguang]

I notice that, thanks.

[shuluoshu]

use car.fhd.config wo have get a 90.12%(easy) in val dateset,so ,I wanna to see the result by using one stage config,thank you.

Hi @yangyongguang , I run the config for 50 epochs, and only achieve 88% for easy 3D object detection(@0.7), I wonder how could you achieve 90% and what about the median and hard result? Thanks!

[yangyongguang]

use car.fhd.config wo have get a 90.12%(easy) in val dateset,so ,I wanna to see the result by using one stage config,thank you.

Hi @yangyongguang , I run the config for 50 epochs, and only achieve 88% for easy 3D object detection(@0.7), I wonder how could you achieve 90% and what about the median and hard result? Thanks!

I have change the model and used the image at the same time. all 3 result is better than the second,but during test in kitti test dataset the median the worser than second, the others is better. and i also run the model without change anything ,the result also worser than the second offer.I remerber the median about 78.2.