evaluate 结果全是0 感谢指导
ksama777 opened this issue · comments
您好，您有空帮我看看全是0是什么情况 麻烦了 谢谢
python3 evaluate_SemanticKITTI.py --dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences_0.06 --predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result/08/predictions --sequence 08
INTERFACE:
Eval What: all
Data: /media/wx/HDD/DQ/semantic_kitti/dataset/sequences_0.06
Predictions: /media/wx/HDD/DQ/RandLA-pytorch-main/result/08/predictions
Sequences: ['08']
Config: /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Limit: None
Opening data config file /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Ignoring xentropy class 0 in IoU evaluation
[IOU EVAL] IGNORE: [0]
[IOU EVAL] INCLUDE: [ 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19]
0
0 0
Evaluating sequences
0it [00:00, ?it/s]
Validation set:
Acc avg 0.000
IoU avg 0.000
IoU class 1 [car] = 0.000
IoU class 2 [bicycle] = 0.000
IoU class 3 [motorcycle] = 0.000
IoU class 4 [truck] = 0.000
IoU class 5 [other-vehicle] = 0.000
IoU class 6 [person] = 0.000
IoU class 7 [bicyclist] = 0.000
IoU class 8 [motorcyclist] = 0.000
IoU class 9 [road] = 0.000
IoU class 10 [parking] = 0.000
IoU class 11 [sidewalk] = 0.000
IoU class 12 [other-ground] = 0.000
IoU class 13 [building] = 0.000
IoU class 14 [fence] = 0.000
IoU class 15 [vegetation] = 0.000
IoU class 16 [trunk] = 0.000
IoU class 17 [terrain] = 0.000
IoU class 18 [pole] = 0.000
IoU class 19 [traffic-sign] = 0.000
0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000
请问prediction中的文件是什么格式的? seq 08 的 prediction是不需要使用--index-to-label的标志的, 不知道您在test的时候是否添加了?
本来是.label格式的 全是0 然后我刚才按您说的 改成全是...npy的了 还是全是0 用的您的预训练的checkpoint
ksama
***@***.***
…------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年9月30日(星期五) 下午5:54
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
请问prediction中的文件是什么格式的? seq 08 的 prediction是不需要使用--index-to-label的标志的, 不知道您在test的时候是否添加了?
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
本来是.label格式的 全是0 然后我刚才按您说的 改成全是.npy的了 还是全是0 用的您的预训练的checkpoint 和我自己的训练测试的checkpoint 也是0
ksama
***@***.***
…------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年9月30日(星期五) 下午5:54
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
请问prediction中的文件是什么格式的? seq 08 的 prediction是不需要使用--index-to-label的标志的, 不知道您在test的时候是否添加了?
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
您好, 我发现您将type设置为了all, 这种场景下--dataset对应的路径应为原数据路径, 在您的场景我猜测应为 /media/wx/HDD/DQ/semantic_kitti/dataset/sequences
如果依然未生效, 我推荐您可以进行逐步排查:
- 确认dataset是否已经使用preprocess处理完成.
- 确认test_Semantickitti.py 正常预测结果, 如可以将预测的.npy读取并查看内容是否为0或有异常, 或在test期间打印概率更新来确认逻辑是否正常更新.
- 确认evaluate操作的对象正常, 如可以检查 处的打印结果是否正常
RandLA-pytorch/evaluate_SemanticKITTI.py
Line 130 in f6e8b40
如果依然有问题也欢迎继续提问
您好
1 我前面的数据预处理 训练 测试 都是按照您的步骤来的，都正常能跑通
--dataset 选择sequences文件夹和sequences_0.06文件夹都不行 另外 eval 选all或者sub 也都不行
2 关于第二步 我通过 import numpy as np
test = np.load('/media/wx/HDD/DQ/RandLA-pytorch-main/result/11/predictions/000000.label')
print(test)
结果是[10 10 10 ... 2 2 2]
不知道这样是否是正确？
3 然后除了全是0的报错 还有您说的第三步的这个报错 我不知道怎么解决
(randlanet) ***@***.***:/media/wx/HDD/DQ/RandLA-pytorch-main$ python3 evaluate_SemanticKITTI.py --dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences --predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result/res/08 --sequence 08 --e all
********************************************************************************
INTERFACE:
Eval What: all
Data: /media/wx/HDD/DQ/semantic_kitti/dataset/sequences
Predictions: /media/wx/HDD/DQ/RandLA-pytorch-main/result/res/08
Sequences: ['08']
Config: /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Limit: None
********************************************************************************
Opening data config file /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Ignoring xentropy class 0 in IoU evaluation
[IOU EVAL] IGNORE: [0]
[IOU EVAL] INCLUDE: [ 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19]
4071
4071 0
Traceback (most recent call last):
File "evaluate_SemanticKITTI.py", line 131, in <module>
assert(len(label_names) == len(pred_names))
AssertionError
(randlanet) ***@***.***:/media/wx/HDD/DQ/RandLA-pytorch-main$ python3 evaluate_SemanticKITTI.py --dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences_0.06 --predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result/res/08 --sequence 08 --e all
********************************************************************************
INTERFACE:
Eval What: all
Data: /media/wx/HDD/DQ/semantic_kitti/dataset/sequences_0.06
Predictions: /media/wx/HDD/DQ/RandLA-pytorch-main/result/res/08
Sequences: ['08']
Config: /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Limit: None
********************************************************************************
Opening data config file /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Ignoring xentropy class 0 in IoU evaluation
[IOU EVAL] IGNORE: [0]
[IOU EVAL] INCLUDE: [ 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19]
0
0 0
Evaluating sequences
0it [00:00, ?it/s]
Validation set:
Acc avg 0.000
IoU avg 0.000
IoU class 1 [car] = 0.000
IoU class 2 [bicycle] = 0.000
IoU class 3 [motorcycle] = 0.000
IoU class 4 [truck] = 0.000
IoU class 5 [other-vehicle] = 0.000
IoU class 6 [person] = 0.000
IoU class 7 [bicyclist] = 0.000
IoU class 8 [motorcyclist] = 0.000
IoU class 9 [road] = 0.000
IoU class 10 [parking] = 0.000
IoU class 11 [sidewalk] = 0.000
IoU class 12 [other-ground] = 0.000
IoU class 13 [building] = 0.000
IoU class 14 [fence] = 0.000
IoU class 15 [vegetation] = 0.000
IoU class 16 [trunk] = 0.000
IoU class 17 [terrain] = 0.000
IoU class 18 [pole] = 0.000
IoU class 19 [traffic-sign] = 0.000
********************************************************************************
0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000
目前卡在这两种报错
ksama
***@***.***
…------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年10月2日(星期天) 中午11:28
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
您好, 我发现您将type设置为了all, 这种场景下--dataset对应的路径应为原数据路径, 在您的场景我猜测应为 /media/wx/HDD/DQ/semantic_kitti/dataset/sequences
如果依然未生效, 我推荐您可以进行逐步排查:
确认dataset是否已经使用preprocess处理完成.
确认test_Semantickitti.py 正常预测结果, 如可以将预测的.npy读取并查看内容是否为0或有异常, 或在test期间打印概率更新来确认逻辑是否正常更新.
确认evaluate操作的对象正常, 如可以检查 https://github.com/charlesChen02/RandLA-pytorch/blob/f6e8b40537f56b0788d6a7dcc09682635631df5f/evaluate_SemanticKITTI.py#L130 处的打印结果是否正常
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
prediction 的路径需要设置成与test相同的路径, 如你设置test目录为res, 那么eval时也只需要设置成res即可, 不需要设置为res/08
python3 evaluate_SemanticKITTI.py --dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences --predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result/res --sequence 08
可以尝试一下这条命令
您好 eval的路径已经和test的一样了 还是这个问题 很头痛 这个4071和0 对不上 导致不能运行
请问可以帮我再看看吗 感谢
(randlanet) ***@***.***:/media/wx/HDD/DQ/RandLA-pytorch-main$ python3 evaluate_SemanticKITTI.py --dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences --predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result --sequence 08
********************************************************************************
INTERFACE:
Eval What: all
Data: /media/wx/HDD/DQ/semantic_kitti/dataset/sequences
Predictions: /media/wx/HDD/DQ/RandLA-pytorch-main/result
Sequences: ['08']
Config: /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Limit: None
********************************************************************************
Opening data config file /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Ignoring xentropy class 0 in IoU evaluation
[IOU EVAL] IGNORE: [0]
[IOU EVAL] INCLUDE: [ 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19]
4071
4071 0
Traceback (most recent call last):
File "evaluate_SemanticKITTI.py", line 131, in <module>
assert(len(label_names) == len(pred_names))
AssertionError
ksama
***@***.***
…------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年10月3日(星期一) 上午6:34
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
python3 evaluate_SemanticKITTI.py --dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences --predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result --sequence 08
可以尝试一下这条命令
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
从报错来看, 就是prediction对应的label文件没有找到
可以大概提供一下result目录下的部分结构吗, 因为我看你之前用的是result/res/08, 这次用的又是result, 不确定你的prediction具体是放在哪个里面的,
简单的说, 如果你需要预测 seq 08, 那么你--prediction 后跟的路径内 应该有 08/prediction 这个子文件夹, 比如 --predictions 后提供路径为 xxx, 那么xxx内路径应包含的结构大概为:
|xxx
|- 08
|- predictions
您好， 我的08的prediction两个路径我都放了 分别为
|result |- 08 |- predictions和|result |- res |- 08 |- predictions
然后我这种的报错是这样的 变成 0 500 了
python3 evaluate_SemanticKITTI.py --dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences --predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result/res --sequence 08 -e sub
********************************************************************************
INTERFACE:
Eval What: sub
Data: /media/wx/HDD/DQ/semantic_kitti/dataset/sequences
Predictions: /media/wx/HDD/DQ/RandLA-pytorch-main/result/res
Sequences: ['08']
Config: /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Limit: None
********************************************************************************
Opening data config file /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Ignoring xentropy class 0 in IoU evaluation
[IOU EVAL] IGNORE: [0]
[IOU EVAL] INCLUDE: [ 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19]
0 500
Traceback (most recent call last):
File "evaluate_SemanticKITTI.py", line 131, in <module>
assert(len(label_names) == len(pred_names))
AssertionError
ksama
***@***.***
…------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年10月4日(星期二) 中午12:01
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
从报错来看, 就是prediction对应的label文件没有找到
可以大概提供一下result目录下的部分结构吗, 因为我看你之前用的是result/res/08, 这次用的又是result, 不确定你的prediction具体是放在哪个里面的,
简单的说, 如果你需要预测 seq 08, 那么你--prediction 后跟的路径内 应该有 08/prediction 这个子文件夹, 比如 --predictions 后提供路径为 xxx, 那么xxx内路径应包含的结构大概为:
|xxx |- 08 |- predictions
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
请不要乱改编译选项... 不方便定位具体问题
作为解决当前问题的方法, 请将prediction对应的目录设置为test时使用all flag的目录, 然后 在evaluate时使用 -e all进行预测,
如果看到报错包含4071 500的话则说明目录选择是正确的
之后请逐步切片并预测valid集中全部的数据, 之前我说的[:500]表明只是预测了前500个数据, 需要反复执行多次获得全部的prediction之后再进行eval
-e all 代表的是对原数据中所有的点进行预测, sub 代表的是对preprocess之后的剩余的点进行预测
我懂了， 那我现在是需要在把之前的[:500] 改成[:4071]然后重新试试吗 我如果不切片的话好像会有内存泄漏的问题
ksama
***@***.***
…------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年10月4日(星期二) 中午12:20
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
请不要乱改编译选项... 不方便定位具体问题
作为解决当前问题的方法, 请将prediction对应的目录设置为test时使用all flag的目录, 然后 在evaluate时使用 -e all进行预测,
如果看到报错包含4071 500的话则说明目录选择是正确的
之后请逐步切片并预测valid集中全部的数据, 之前我说的[:500]表明只是预测了前500个数据, 需要反复执行多次获得全部的prediction之后再进行eval
-e all 代表的是对原数据中所有的点进行预测, sub 代表的是对preprocess之后的剩余的点进行预测
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
如果内存不够, 可以跑多次test, 一次跑500个, 这样每次就不会超内存, 比如test第一次切[:500], 第二次test切[500:1000], 以此类推
已经成功运行了 非常感谢您的指导！！！ 请问08的 这些数据是正常的吗 因为我看有的是0.000
Evaluating sequences
100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 4071/4071 [02:15<00:00, 30.00it/s]
Validation set:
Acc avg 0.040
IoU avg 0.010
IoU class 1 [car] = 0.040
IoU class 2 [bicycle] = 0.000
IoU class 3 [motorcycle] = 0.000
IoU class 4 [truck] = 0.000
IoU class 5 [other-vehicle] = 0.000
IoU class 6 [person] = 0.001
IoU class 7 [bicyclist] = 0.000
IoU class 8 [motorcyclist] = 0.000
IoU class 9 [road] = 0.056
IoU class 10 [parking] = 0.000
IoU class 11 [sidewalk] = 0.023
IoU class 12 [other-ground] = 0.000
IoU class 13 [building] = 0.000
IoU class 14 [fence] = 0.031
IoU class 15 [vegetation] = 0.035
IoU class 16 [trunk] = 0.002
IoU class 17 [terrain] = 0.000
IoU class 18 [pole] = 0.004
IoU class 19 [traffic-sign] = 0.000
********************************************************************************
0.040,0.000,0.000,0.000,0.000,0.001,0.000,0.000,0.056,0.000,0.023,0.000,0.000,0.031,0.035,0.002,0.000,0.004,0.000,0.010,0.040
ksama
***@***.***
…------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年10月4日(星期二) 中午12:29
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
如果内存不够, 可以跑多次test, 一次跑500个, 这样每次就不会超内存, 比如test第一次切[:500], 第二次test切[500:1000], 以此类推
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
不太正常, 请问您test时的命令是什么样的
您好， 我运行visualize的时候 这个bug是啥原因？ 运行visualize的时候有啥要注意的吗？
我这样设置路径是对的吗？ 我在test的时候 给它放在了result/vis 里面 然后还 --index_to_label 了 我的result/vis 里面就是08/predictions 里面都是.label的文件
(randlanet) ***@***.***:/media/wx/HDD/DQ/RandLA-pytorch-main$ python3 visualize_SemanticKITTI.py --dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences --predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result/vis/08 --sequence 08
********************************************************************************
INTERFACE:
Dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences
Config /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Sequence 08
Predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result/vis/08
********************************************************************************
Opening config file /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Sequence folder exists! Using sequence from /media/wx/HDD/DQ/semantic_kitti/dataset/sequences/08/velodyne
Labels folder exists! Using labels from /media/wx/HDD/DQ/RandLA-pytorch-main/result/vis/08/predictions
Using semantics in visualizer
Traceback (most recent call last):
File "visualize_SemanticKITTI.py", line 98, in <module>
semantics=True, instances=False)
File "/media/wx/HDD/DQ/RandLA-pytorch-main/utils/semkitti_vis/laserscanvis.py", line 28, in __init__
self.update_scan()
File "/media/wx/HDD/DQ/RandLA-pytorch-main/utils/semkitti_vis/laserscanvis.py", line 108, in update_scan
self.scan.open_scan(self.scan_names[self.offset])
File "/media/wx/HDD/DQ/RandLA-pytorch-main/utils/semkitti_vis/laserscan.py", line 68, in open_scan
raise RuntimeError("Filename extension is not valid scan file.")
RuntimeError: Filename extension is not valid scan file.
ksama
***@***.***
…------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年10月4日(星期二) 中午12:29
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
如果内存不够, 可以跑多次test, 一次跑500个, 这样每次就不会超内存, 比如test第一次切[:500], 第二次test切[500:1000], 以此类推
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
您好我test的命令是 python3 test_SemanticKITTI.py --test_id 08 --result_dir /media/wx/HDD/DQ/RandLA-pytorch-main/result/res
然后请问除了08， 11-21的那些是怎么eval? 是加上--index_to_label的结果（.label文件） 放到sequences 文件里面吗？
我这样操作 数字又对不上了 比如12的eval 就是0 1061
ksama
***@***.***
…------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年10月5日(星期三) 凌晨5:22
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
不太正常, 请问您test时的命令是什么样的
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
您好，我把11-21的.label文件上传网站来eval 结果miou只有1.1 macc4.8 请问这个可能是什么情况。。。。
accuracy_mean: 4.759975250038975
iou_bicycle: 0.03359058233529147
iou_bicyclist: 0.0
iou_building: 0.05596576902588247
iou_car: 2.1014818305831966
iou_fence: 1.7276567002605552
iou_mean: 1.1079832610845668
iou_motorcycle: 0.004439356164924579
iou_motorcyclist: 0.0
iou_moving-bicyclist: 0.0
iou_moving-car: 0.0
iou_moving-motorcyclist: 0.0
iou_moving-other-vehicle: 0.0
iou_moving-person: 0.0
iou_moving-truck: 0.0
iou_other-ground: 0.08546182852475125
iou_other-vehicle: 0.05033995477673206
iou_parking: 3.9664307493264253
iou_person: 0.029447094915394975
iou_pole: 0.5008478885181411
iou_road: 3.5254680546235866
iou_sidewalk: 4.26834879554106
iou_terrain: 5.252210887772718
iou_traffic-sign: 0.21301859334245396
iou_truck: 0.436186113868868
iou_trunk: 0.48656358026659363
iou_vegetation: 4.962123747267589
ksama
***@***.***
…------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年10月9日(星期天) 上午8:16
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
#3 (comment)
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
您好， 我运行visualize的时候 这个bug是啥原因？ 运行visualize的时候有啥要注意的吗？
我这样设置路径是对的吧 我在test的时候 给它放在了result/vis 里面 然后还 --index_to_label 了
(randlanet) ***@***.***:/media/wx/HDD/DQ/RandLA-pytorch-main$ python3 visualize_SemanticKITTI.py --dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences --predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result/vis/08 --sequence 08
********************************************************************************
INTERFACE:
Dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences
Config /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Sequence 08
Predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result/vis/08
********************************************************************************
Opening config file /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Sequence folder exists! Using sequence from /media/wx/HDD/DQ/semantic_kitti/dataset/sequences/08/velodyne
Labels folder exists! Using labels from /media/wx/HDD/DQ/RandLA-pytorch-main/result/vis/08/predictions
Using semantics in visualizer
Traceback (most recent call last):
File "visualize_SemanticKITTI.py", line 98, in <module>
semantics=True, instances=False)
File "/media/wx/HDD/DQ/RandLA-pytorch-main/utils/semkitti_vis/laserscanvis.py", line 28, in __init__
self.update_scan()
File "/media/wx/HDD/DQ/RandLA-pytorch-main/utils/semkitti_vis/laserscanvis.py", line 108, in update_scan
self.scan.open_scan(self.scan_names[self.offset])
File "/media/wx/HDD/DQ/RandLA-pytorch-main/utils/semkitti_vis/laserscan.py", line 68, in open_scan
raise RuntimeError("Filename extension is not valid scan file.")
RuntimeError: Filename extension is not valid scan file.
ksama
***@***.***
…------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年10月4日(星期二) 中午12:29
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
如果内存不够, 可以跑多次test, 一次跑500个, 这样每次就不会超内存, 比如test第一次切[:500], 第二次test切[500:1000], 以此类推
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
您好， 我的08的prediction两个路径我都放了 分别为
|result |- 08 |- predictions和|result |- res |- 08 |- predictions
然后我这种的报错是这样的 变成 0 500 了
python3 evaluate_SemanticKITTI.py --dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences --predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result/res --sequence 08 -e sub
********************************************************************************
INTERFACE:
Eval What: sub
Data: /media/wx/HDD/DQ/semantic_kitti/dataset/sequences
Predictions: /media/wx/HDD/DQ/RandLA-pytorch-main/result/res
Sequences: ['08']
Config: /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Limit: None
********************************************************************************
Opening data config file /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Ignoring xentropy class 0 in IoU evaluation
[IOU EVAL] IGNORE: [0]
[IOU EVAL] INCLUDE: [ 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19]
0 500
Traceback (most recent call last):
File "evaluate_SemanticKITTI.py", line 131, in <module>
assert(len(label_names) == len(pred_names))
AssertionError
如果不加上-e sub 就是这样 4071 500
(randlanet) ***@***.***:/media/wx/HDD/DQ/RandLA-pytorch-main$ python3 evaluate_SemanticKITTI.py --dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences --predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result/res --sequence 08
********************************************************************************
INTERFACE:
Eval What: all
Data: /media/wx/HDD/DQ/semantic_kitti/dataset/sequences
Predictions: /media/wx/HDD/DQ/RandLA-pytorch-main/result/res
Sequences: ['08']
Config: /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Limit: None
********************************************************************************
Opening data config file /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Ignoring xentropy class 0 in IoU evaluation
[IOU EVAL] IGNORE: [0]
[IOU EVAL] INCLUDE: [ 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19]
4071
4071 500
Traceback (most recent call last):
File "evaluate_SemanticKITTI.py", line 131, in <module>
assert(len(label_names) == len(pred_names))
AssertionError
ksama
***@***.***
…------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年10月4日(星期二) 中午12:01
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
从报错来看, 就是prediction对应的label文件没有找到
可以大概提供一下result目录下的部分结构吗, 因为我看你之前用的是result/res/08, 这次用的又是result, 不确定你的prediction具体是放在哪个里面的,
简单的说, 如果你需要预测 seq 08, 那么你--prediction 后跟的路径内 应该有 08/prediction 这个子文件夹, 比如 --predictions 后提供路径为 xxx, 那么xxx内路径应包含的结构大概为:
|xxx |- 08 |- predictions
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
您好 eval的路径已经和test的一样了 还是这个问题 很头痛 这个4071和0 对不上 导致不能运行
(randlanet) ***@***.***:/media/wx/HDD/DQ/RandLA-pytorch-main$ python3 evaluate_SemanticKITTI.py --dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences --predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result --sequence 08
********************************************************************************
INTERFACE:
Eval What: all
Data: /media/wx/HDD/DQ/semantic_kitti/dataset/sequences
Predictions: /media/wx/HDD/DQ/RandLA-pytorch-main/result
Sequences: ['08']
Config: /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Limit: None
********************************************************************************
Opening data config file /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Ignoring xentropy class 0 in IoU evaluation
[IOU EVAL] IGNORE: [0]
[IOU EVAL] INCLUDE: [ 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19]
4071
4071 0
Traceback (most recent call last):
File "evaluate_SemanticKITTI.py", line 131, in <module>
assert(len(label_names) == len(pred_names))
AssertionError
ksama
***@***.***
…------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年10月3日(星期一) 上午6:34
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
python3 evaluate_SemanticKITTI.py --dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences --predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result --sequence 08
可以尝试一下这条命令
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
在此之前，想问一下您使用提供的模型测试seq08得到了与主页写的相似的预测结果了吗，是否确认整个prediction流程没出什么差错？
…On Mon, 10 Oct 2022 at 2:23 pm, ksama777 ***@***.***> wrote:
您好，我把11-21的.label文件上传网站来eval 结果miou只有1.1 macc4.8
请问这个可能是什么情况。。。。
accuracy_mean: 4.759975250038975
iou_bicycle: 0.03359058233529147
iou_bicyclist: 0.0
iou_building: 0.05596576902588247
iou_car: 2.1014818305831966
iou_fence: 1.7276567002605552
iou_mean: 1.1079832610845668
iou_motorcycle: 0.004439356164924579
iou_motorcyclist: 0.0
iou_moving-bicyclist: 0.0
iou_moving-car: 0.0
iou_moving-motorcyclist: 0.0
iou_moving-other-vehicle: 0.0
iou_moving-person: 0.0
iou_moving-truck: 0.0
iou_other-ground: 0.08546182852475125
iou_other-vehicle: 0.05033995477673206
iou_parking: 3.9664307493264253
iou_person: 0.029447094915394975
iou_pole: 0.5008478885181411
iou_road: 3.5254680546235866
iou_sidewalk: 4.26834879554106
iou_terrain: 5.252210887772718
iou_traffic-sign: 0.21301859334245396
iou_truck: 0.436186113868868
iou_trunk: 0.48656358026659363
iou_vegetation: 4.962123747267589
ksama
***@***.***
------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年10月9日(星期天) 上午8:16
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
#3 (comment)
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID:
***@***.***>
—
Reply to this email directly, view it on GitHub
<#4 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AIZALRGS47BJE3PEWTZK6CTWCOD25ANCNFSM6AAAAAAQZRLRPA>
.
You are receiving this because you commented.Message ID:
***@***.***>
您好 08的也是1点多的miou 整个test4071个 我500个500个一test 得到的都是.npy（08的） 然后对应的sequence文件夹 （之前您教的） 顺利运行 但是结果就是不正常
会不会是我train用的valbatchsize过小？（我调成2）
ksama
***@***.***
…------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年10月11日(星期二) 下午4:15
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
在此之前，想问一下您使用提供的模型测试seq08得到了与主页写的相似的预测结果了吗，是否确认整个prediction流程没出什么差错？
On Mon, 10 Oct 2022 at 2:23 pm, ksama777 ***@***.***> wrote:
> 您好，我把11-21的.label文件上传网站来eval&nbsp;结果miou只有1.1&nbsp; macc4.8
> 请问这个可能是什么情况。。。。
>
>
> accuracy_mean: 4.759975250038975
> iou_bicycle: 0.03359058233529147
> iou_bicyclist: 0.0
> iou_building: 0.05596576902588247
> iou_car: 2.1014818305831966
> iou_fence: 1.7276567002605552
> iou_mean: 1.1079832610845668
> iou_motorcycle: 0.004439356164924579
> iou_motorcyclist: 0.0
> iou_moving-bicyclist: 0.0
> iou_moving-car: 0.0
> iou_moving-motorcyclist: 0.0
> iou_moving-other-vehicle: 0.0
> iou_moving-person: 0.0
> iou_moving-truck: 0.0
> iou_other-ground: 0.08546182852475125
> iou_other-vehicle: 0.05033995477673206
> iou_parking: 3.9664307493264253
> iou_person: 0.029447094915394975
> iou_pole: 0.5008478885181411
> iou_road: 3.5254680546235866
> iou_sidewalk: 4.26834879554106
> iou_terrain: 5.252210887772718
> iou_traffic-sign: 0.21301859334245396
> iou_truck: 0.436186113868868
> iou_trunk: 0.48656358026659363
> iou_vegetation: 4.962123747267589
>
>
>
>
>
>
>
>
>
> ksama
> ***@***.***
>
>
>
> &nbsp;
>
>
>
>
> ------------------&nbsp;原始邮件&nbsp;------------------
> 发件人: "charlesChen02/RandLA-pytorch" ***@***.***&gt;;
> 发送时间:&nbsp;2022年10月9日(星期天) 上午8:16
> ***@***.***&gt;;
> ***@***.******@***.***&gt;;
> 主题:&nbsp;Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
>
>
>
>
>
>
> #3 (comment)
>
> —
> Reply to this email directly, view it on GitHub, or unsubscribe.
> You are receiving this because you authored the thread.Message ID:
> ***@***.***&gt;
>
> —
> Reply to this email directly, view it on GitHub
> <#4 (comment)>,
> or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/AIZALRGS47BJE3PEWTZK6CTWCOD25ANCNFSM6AAAAAAQZRLRPA>
> .
> You are receiving this because you commented.Message ID:
> ***@***.***>
>
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
如果您有其他问题, 麻烦单独开一个issue, 具体描述一下问题, 所操作的命令, 需要求助的点, 如遇到报错无法理解也欢迎提出issue, 目前看来mIoU为0的问题解决了, 我先关闭这个issue了
您好
1 我前面的数据预处理 训练 测试 都是按照您的步骤来的，都正常能跑通
--dataset 选择sequences文件夹和sequences_0.06文件夹都不行 另外 eval 选all或者sub 也都不行
2 关于第二步 我通过 import numpy as np
test = np.load('/media/wx/HDD/DQ/RandLA-pytorch-main/result/res/08/predictions/000000.npy')
print(test)
结果是[10 10 10 ... 2 2 2]
不知道这样是否是正确？
3 然后除了全是0的报错 还有您说的第三步的这个报错 我不知道怎么解决
(randlanet) ***@***.***:/media/wx/HDD/DQ/RandLA-pytorch-main$ python3 evaluate_SemanticKITTI.py --dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences --predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result/res/08 --sequence 08 --e all
********************************************************************************
INTERFACE:
Eval What: all
Data: /media/wx/HDD/DQ/semantic_kitti/dataset/sequences
Predictions: /media/wx/HDD/DQ/RandLA-pytorch-main/result/res/08
Sequences: ['08']
Config: /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Limit: None
********************************************************************************
Opening data config file /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Ignoring xentropy class 0 in IoU evaluation
[IOU EVAL] IGNORE: [0]
[IOU EVAL] INCLUDE: [ 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19]
4071
4071 0
Traceback (most recent call last):
File "evaluate_SemanticKITTI.py", line 131, in <module>
assert(len(label_names) == len(pred_names))
AssertionError
(randlanet) ***@***.***:/media/wx/HDD/DQ/RandLA-pytorch-main$ python3 evaluate_SemanticKITTI.py --dataset /media/wx/HDD/DQ/semantic_kitti/dataset/sequences_0.06 --predictions /media/wx/HDD/DQ/RandLA-pytorch-main/result/res/08 --sequence 08 --e all
********************************************************************************
INTERFACE:
Eval What: all
Data: /media/wx/HDD/DQ/semantic_kitti/dataset/sequences_0.06
Predictions: /media/wx/HDD/DQ/RandLA-pytorch-main/result/res/08
Sequences: ['08']
Config: /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Limit: None
********************************************************************************
Opening data config file /media/wx/HDD/DQ/RandLA-pytorch-main/utils/semantic-kitti.yaml
Ignoring xentropy class 0 in IoU evaluation
[IOU EVAL] IGNORE: [0]
[IOU EVAL] INCLUDE: [ 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19]
0
0 0
Evaluating sequences
0it [00:00, ?it/s]
Validation set:
Acc avg 0.000
IoU avg 0.000
IoU class 1 [car] = 0.000
IoU class 2 [bicycle] = 0.000
IoU class 3 [motorcycle] = 0.000
IoU class 4 [truck] = 0.000
IoU class 5 [other-vehicle] = 0.000
IoU class 6 [person] = 0.000
IoU class 7 [bicyclist] = 0.000
IoU class 8 [motorcyclist] = 0.000
IoU class 9 [road] = 0.000
IoU class 10 [parking] = 0.000
IoU class 11 [sidewalk] = 0.000
IoU class 12 [other-ground] = 0.000
IoU class 13 [building] = 0.000
IoU class 14 [fence] = 0.000
IoU class 15 [vegetation] = 0.000
IoU class 16 [trunk] = 0.000
IoU class 17 [terrain] = 0.000
IoU class 18 [pole] = 0.000
IoU class 19 [traffic-sign] = 0.000
********************************************************************************
0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000,0.000
目前卡在这两种报错
ksama
***@***.***
…------------------ 原始邮件 ------------------
发件人: "charlesChen02/RandLA-pytorch" ***@***.***>;
发送时间: 2022年10月2日(星期天) 中午11:28
***@***.***>;
***@***.******@***.***>;
主题: Re: [charlesChen02/RandLA-pytorch] evaluate 结果全是0 感谢指导 (Issue #4)
您好, 我发现您将type设置为了all, 这种场景下--dataset对应的路径应为原数据路径, 在您的场景我猜测应为 /media/wx/HDD/DQ/semantic_kitti/dataset/sequences
如果依然未生效, 我推荐您可以进行逐步排查:
确认dataset是否已经使用preprocess处理完成.
确认test_Semantickitti.py 正常预测结果, 如可以将预测的.npy读取并查看内容是否为0或有异常, 或在test期间打印概率更新来确认逻辑是否正常更新.
确认evaluate操作的对象正常, 如可以检查 https://github.com/charlesChen02/RandLA-pytorch/blob/f6e8b40537f56b0788d6a7dcc09682635631df5f/evaluate_SemanticKITTI.py#L130 处的打印结果是否正常
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>