question about pretrained model?

[Zhangjyhhh]

could you please tell me the difference between your provide pretrain model and the rangenet++ trained mode? here is my trained model , the backone file seems different from yours

[G12311231]

你能告诉我你提供的预训练模型和 rangenet++ 训练模式之间的区别吗？ 这是我训练的模型，backone 文件似乎与你的不同

Hi, I would like to ask how you get the training model. Is it necessary to run lidar-bonnetal? The tensorrt version required is too old. The tensorrt I installed is 8.5.1.7. I would like to know your solution.Thank you!

[Zhangjyhhh]

你能告诉我你提供的预训练模型和 rangenet++ 训练模式之间的区别吗？ 这是我训练的模型，backone 文件似乎与你的不同

Hi, I would like to ask how you get the training model. Is it necessary to run lidar-bonnetal? The tensorrt version required is too old. The tensorrt I installed is 8.5.1.7. I would like to know your solution.Thank you!

yes, I get my training model by running lidar-bonnetal, but the author provide traning model, so you don't have to run lidar-bonnetal. there is a solution for higher tensorrt, you could reference this linkhttps://github.com/PRBonn/rangenet_lib/issues/9

[G12311231]

Thank you very much for your answer! It means that when I want to use a custom dataset to train the network, I need to run lidar-bonnetal successfully. Otherwise, I can directly use the model trained by the author to run, refer to rangenet_lib, right?

[Zhangjyhhh]

Thank you very much for your answer! It means that when I want to use a custom dataset to train the network, I need to run lidar-bonnetal successfully. Otherwise, I can directly use the model trained by the author to run, refer to rangenet_lib, right?

right!!!