search

lucastabelini / PolyLaneNet

Code for the paper entitled "PolyLaneNet: Lane Estimation via Deep Polynomial Regression" (ICPR 2020)
https://arxiv.org/abs/2004.10924
MIT License
295 stars 74 forks source link

Confidence Scores #60

Closed iremonur closed 2 years ago

iremonur commented 2 years ago

Hi, the confidence scores in the log.txt files in which located in the experiment files (via the Drive link provided) are extremely low (in the range of 0.0 - 0.0005). Is this an expected result? Should not be the confidence scores higher while the loss decreases to approximately 0.005 at the end of the training?

lucastabelini commented 2 years ago

The predicted confidence scores are not in the provided log.txt files. The conf value shown is the loss component related to the confidence score.

iremonur commented 2 years ago

Thank you for your response, so what information does the conf score shown in the log.txt files carry? Is this related to training performance? And also, I want the model to classify the lanes (solid, dashed, double). Is it possible, and if so, how? I would be glad if you help me with this.

lucastabelini commented 2 years ago

Thank you for your response, so what information does the conf score shown in the log.txt files carry? Is this related to training performance?

As stated previously, the conf value shown is the loss component related to the confidence score.

And also, I want the model to classify the lanes (solid, dashed, double). Is it possible, and if so, how?

Yes, it is possible. The elas.py file is an example on how to load a dataset with classification annotations. You can use it as a base to provide a data loader for your dataset.