how to get prediction.json from the results predicted from tusimple for comparing with the labels of tusimple? about robust-lane-detection HOT 3 OPEN

We compute the accuracy, precision, recall, f1 by comparing the detected results and the ground truth. The codes can be found in the function `evaluate_model' in the test.py file.

from robust-lane-detection.

accuracy, precision, recall, f1 is different metric with tusimple metric, in your paper, you provide the result calculated according to the rule of tusimple, first is to get json file from the prediction, second is to calculate acc, fp, fn, l want to known the detailed process of calculating x coordinates from the prediction images. look forward reply, thank you very much

from robust-lane-detection.

Same question! How to evaluate with tusimple metric. We can only get binary segmentation results without lane instance discrimination.

from robust-lane-detection.