E2E-ERFNet
The re-implementation of <End-to-End Lane Marker Detection via Row-wise Classification>
The original paper implemented CULane dataset for 74% F1 score. We re-implemented the models for 67.2%. We believe that there are some differences in the implementation details compared with the original text. (We didn't add dropout operation to the network because it dropped 2% F1 score.)
result
------------Configuration--------- anno_dir: /mnt/HD/dataset/CULane/ detect_dir: ./result/culane_eval_tmp/ im_dir: /mnt/HD/dataset/CULane/ list_im_file: /mnt/HD/dataset/CULane/list/test_split/test0_normal.txt width_lane: 30 iou_threshold: 0.5 im_width: 1640 im_height: 590
Evaluating the results... tp: 28271 fp: 4438 fn: 4506 finished process file precision: 0.864319 recall: 0.862526 Fmeasure: 0.863421
------------Configuration--------- anno_dir: /mnt/HD/dataset/CULane/ detect_dir: ./result/culane_eval_tmp/ im_dir: /mnt/HD/dataset/CULane/ list_im_file: /mnt/HD/dataset/CULane/list/test_split/test1_crowd.txt width_lane: 30 iou_threshold: 0.5 im_width: 1640 im_height: 590
Evaluating the results... tp: 18681 fp: 9001 fn: 9322 finished process file precision: 0.674843 recall: 0.667107 Fmeasure: 0.670953
------------Configuration--------- anno_dir: /mnt/HD/dataset/CULane/ detect_dir: ./result/culane_eval_tmp/ im_dir: /mnt/HD/dataset/CULane/ list_im_file: /mnt/HD/dataset/CULane/list/test_split/test2_hlight.txt width_lane: 30 iou_threshold: 0.5 im_width: 1640 im_height: 590
Evaluating the results... tp: 963 fp: 647 fn: 722 finished process file precision: 0.598137 recall: 0.571513 Fmeasure: 0.584522
------------Configuration--------- anno_dir: /mnt/HD/dataset/CULane/ detect_dir: ./result/culane_eval_tmp/ im_dir: /mnt/HD/dataset/CULane/ list_im_file: /mnt/HD/dataset/CULane/list/test_split/test3_shadow.txt width_lane: 30 iou_threshold: 0.5 im_width: 1640 im_height: 590
Evaluating the results... tp: 1756 fp: 1118 fn: 1120 finished process file precision: 0.610995 recall: 0.61057 Fmeasure: 0.610783
------------Configuration--------- anno_dir: /mnt/HD/dataset/CULane/ detect_dir: ./result/culane_eval_tmp/ im_dir: /mnt/HD/dataset/CULane/ list_im_file: /mnt/HD/dataset/CULane/list/test_split/test4_noline.txt width_lane: 30 iou_threshold: 0.5 im_width: 1640 im_height: 590
Evaluating the results... tp: 5544 fp: 7611 fn: 8477 finished process file precision: 0.421437 recall: 0.395407 Fmeasure: 0.408007
------------Configuration--------- anno_dir: /mnt/HD/dataset/CULane/ detect_dir: ./result/culane_eval_tmp/ im_dir: /mnt/HD/dataset/CULane/ list_im_file: /mnt/HD/dataset/CULane/list/test_split/test5_arrow.txt width_lane: 30 iou_threshold: 0.5 im_width: 1640 im_height: 590
Evaluating the results... tp: 2492 fp: 592 fn: 690 finished process file precision: 0.808042 recall: 0.783155 Fmeasure: 0.795404
------------Configuration--------- anno_dir: /mnt/HD/dataset/CULane/ detect_dir: ./result/culane_eval_tmp/ im_dir: /mnt/HD/dataset/CULane/ list_im_file: /mnt/HD/dataset/CULane/list/test_split/test6_curve.txt width_lane: 30 iou_threshold: 0.5 im_width: 1640 im_height: 590
Evaluating the results... tp: 707 fp: 438 fn: 605 finished process file precision: 0.617467 recall: 0.538872 Fmeasure: 0.575499
------------Configuration--------- anno_dir: /mnt/HD/dataset/CULane/ detect_dir: ./result/culane_eval_tmp/ im_dir: /mnt/HD/dataset/CULane/ list_im_file: /mnt/HD/dataset/CULane/list/test_split/test7_cross.txt width_lane: 30 iou_threshold: 0.5 im_width: 1640 im_height: 590
Evaluating the results... tp: 0 fp: 2841 fn: 0 no ground truth positive finished process file precision: 0 recall: -1 Fmeasure: 0
------------Configuration--------- anno_dir: /mnt/HD/dataset/CULane/ detect_dir: ./result/culane_eval_tmp/ im_dir: /mnt/HD/dataset/CULane/ list_im_file: /mnt/HD/dataset/CULane/list/test_split/test8_night.txt width_lane: 30 iou_threshold: 0.5 im_width: 1640 im_height: 590
Evaluating the results... tp: 12424 fp: 8545 fn: 8606 finished process file precision: 0.592494 recall: 0.590775 Fmeasure: 0.591633
res_normal 0.863421 res_crowd 0.670953 res_night 0.591633 res_noline 0.408007 res_shadow 0.610783 res_arrow 0.795404 res_hlight 0.584522 res_curve 0.575499 res_cross 0.0 0.6715934678011898
Thanks
Some codes implemented by Ultra-fast-laneNet, we appreciated for their works.