search

MaybeShewill-CV / lanenet-lane-detection

Unofficial implemention of lanenet model for real time lane detection
Apache License 2.0
2.36k stars 886 forks source link

Should I resize the pictures in my training set to the same size? #112

Closed allensu0314 closed 5 years ago

allensu0314 commented 5 years ago

Hi, I noticed that the h-sample is all the same for any item in tusimple dataset json file. Now I want to train the net in my datasets, but the images in my set is in various scale. Should I resize them to the same size? Thank you for your code, looking forward to your reply : )

MaybeShewill-CV commented 5 years ago

@Chen-Su Yep all the training samples should be rescaled to the same size:)

allensu0314 commented 5 years ago

@MaybeShewill-CV Got it, thank you again : )

MaybeShewill-CV commented 5 years ago

@Chen-Su Welcome:)

yinhai86924 commented 5 years ago

您好！我看了具体的图森车道线数据集制作格式，，但是图森公司标注车道线时，垂直取统一的每隔10个像素，进行车道线标注，"h_samples": [120, 130, 140, 150, 160, 170, 180, 190, 200, 210, 220, 230, 240, 250, 260, 270, 280, 290, 300, 310, 320, 330, 340, 350, 360, 370, 380, 390, 400, 410, 420, 430, 440, 450, 460, 470, 480, 490, 500, 510, 520, 530, 540, 550, 560, 570, 580, 590, 600, 610, 620, 630, 640, 650, 660, 670, 680, 690, 700, 710]，想制作自己的数据集，对于我能不能在标注点的垂直方向取每隔20个像素 进行车道线标注。用于lannenet进行训练。

MaybeShewill-CV commented 5 years ago

@yinhai86924 标注问题我觉得每隔20个像素也是没有问题的，但是如果点太稀疏的话可能不能用polyline直接连接可能需要做曲线拟合:)