I want to add new class by own Livox data. How can I train?

applededipan commented 3 years ago

I also want to know how...

YmanChris commented 3 years ago

I also want to know how... I have solved the problem today, it worked. You can use any network code(Mine is CIA-SSD) with KITTI Dataset, you should change the input format with livox data type.

applededipan commented 3 years ago

I also want to know how... I have solved the problem today, it worked. You can use any network code(Mine is CIA-SSD) with KITTI Dataset, you should change the input format with livox data type.

THANKS! Can you tell me your wechat? i want to ask some questions. now i can run detection on my nvidia agx xavier , but i takes about 100ms to detect per frame. i think it may runs faster, but i do not know how to do.

YmanChris commented 3 years ago

I also want to know how... I have solved the problem today, it worked. You can use any network code(Mine is CIA-SSD) with KITTI Dataset, you should change the input format with livox data type.

THANKS! Can you tell me your wechat? i want to ask some questions. now i can run detection on my nvidia agx xavier , but i takes about 100ms to detect per frame. i think it may runs faster, but i do not know how to do.

Sent by Email.

www04576 commented 2 years ago

hi @applededipan , I also run livox_detection on nvidia agx xavier, but the detection time per frame is about 1000ms. could you please tell me the version of tensorflow you use? and did you accelerate the detection process by C++? thanks a lot!

lockkedddd commented 10 months ago

I also want to know how... I have solved the problem today, it worked. You can use any network code(Mine is CIA-SSD) with KITTI Dataset, you should change the input format with livox data type.

THANKS! Can you tell me your wechat? i want to ask some questions. now i can run detection on my nvidia agx xavier , but i takes about 100ms to detect per frame. i think it may runs faster, but i do not know how to do.

I would like to run the package I recorded on my computer, using the MID70 radar. How can I train my model or send it to me via WeChat? Thank you very much for your reply

lockkedddd commented 10 months ago

hi @applededipan , I also run livox_detection on nvidia agx xavier, but the detection time per frame is about 1000ms. could you please tell me the version of tensorflow you use? and did you accelerate the detection process by C++? thanks a lot! Hello, I have a few questions about how to training my own model. Would it be convenient to add a WeChat account

Double-zh commented 4 months ago

hi @applededipan , I also run livox_detection on nvidia agx xavier, but the detection time per frame is about 1000ms. could you please tell me the version of tensorflow you use? and did you accelerate the detection process by C++? thanks a lot! Hello, I have a few questions about how to training my own model. Would it be convenient to add a WeChat account

你好啊，请问你有遇到下面的情况吗 ERROR] [1720425207.019895]: bad callback: <bound method ros_demo.online_inference of <main.ros_demo object at 0x7f23a8ecedf0>> Traceback (most recent call last): File "/opt/ros/melodic/lib/python2.7/dist-packages/rospy/topics.py", line 750, in _invoke_callback cb(msg) File "test_ros.py", line 158, in online_inference data_dict = self.receive_from_ros(msg) File "test_ros.py", line 116, in receive_from_ros points_list[:, 3] = copy.deepcopy(np.float32(pc['intensity'])) ValueError: no field of name intensity 感觉是数据的问题，你的rosbag play [bag path]这个数据请问是怎么来的，采用mid70录制就好了嘛

www04576 commented 4 months ago

你好啊，请问你有遇到下面的情况吗 ERROR] [1720425207.019895]: bad callback: <bound method ros_demo.online_inference of <main.ros_demo object at 0x7f23a8ecedf0>> Traceback (most recent call last): File "/opt/ros/melodic/lib/python2.7/dist-packages/rospy/topics.py", line 750, in _invoke_callback cb(msg) File "test_ros.py", line 158, in online_inference data_dict = self.receive_from_ros(msg) File "test_ros.py", line 116, in receive_from_ros points_list[:, 3] = copy.deepcopy(np.float32(pc['intensity'])) ValueError: no field of name intensity 感觉是数据的问题，你的rosbag play [bag path]这个数据请问是怎么来的，采用mid70录制就好了嘛

Hi @Double-zh. It seems to be a data problem. The point cloud has no intensity. There is a rosbag under the v1.1 branch of this repo. You can play it to test it.

lockkedddd commented 4 months ago

你好，我没有遇到过这个问题，我的数据是之前自己录制的，使用的是mid-70，因为mid-70的反射率也是一个重要指标，建议你查看下自己的数据是否包含这个参数。

------------------ 原始邮件 ------------------ 发件人: "Livox-SDK/livox_detection" @.>; 发送时间: 2024年7月8日(星期一) 下午3:55 @.>; @.**@.>; 主题: Re: [Livox-SDK/livox_detection] I want to add new class by own Livox data. How can I train? (#21)

hi @applededipan , I also run livox_detection on nvidia agx xavier, but the detection time per frame is about 1000ms. could you please tell me the version of tensorflow you use? and did you accelerate the detection process by C++? thanks a lot! Hello, I have a few questions about how to training my own model. Would it be convenient to add a WeChat account

你好啊，请问你有遇到下面的情况吗 ERROR] [1720425207.019895]: bad callback: <bound method ros_demo.online_inference of <main.ros_demo object at 0x7f23a8ecedf0>> Traceback (most recent call last): File "/opt/ros/melodic/lib/python2.7/dist-packages/rospy/topics.py", line 750, in _invoke_callback cb(msg) File "test_ros.py", line 158, in online_inference data_dict = self.receive_from_ros(msg) File "test_ros.py", line 116, in receive_from_ros points_list[:, 3] = copy.deepcopy(np.float32(pc['intensity'])) ValueError: no field of name intensity 感觉是数据的问题，你的rosbag play [bag path]这个数据请问是怎么来的，采用mid70录制就好了嘛

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

lockkedddd commented 4 months ago

我采用的是mid-70录制的，建议你查看自己的数据有没有包含 intensity。我最后没有使用livox_detection，而是使用的是PointNet++来做语义分割。

------------------ 原始邮件 ------------------ 发件人: "Livox-SDK/livox_detection" @.>; 发送时间: 2024年7月8日(星期一) 晚上6:14 @.>; @.**@.>; 主题: Re: [Livox-SDK/livox_detection] I want to add new class by own Livox data. How can I train? (#21)

你好啊，请问你有遇到下面的情况吗 ERROR] [1720425207.019895]: bad callback: <bound method ros_demo.online_inference of <main.ros_demo object at 0x7f23a8ecedf0>> Traceback (most recent call last): File "/opt/ros/melodic/lib/python2.7/dist-packages/rospy/topics.py", line 750, in _invoke_callback cb(msg) File "test_ros.py", line 158, in online_inference data_dict = self.receive_from_ros(msg) File "test_ros.py", line 116, in receive_from_ros points_list[:, 3] = copy.deepcopy(np.float32(pc['intensity'])) ValueError: no field of name intensity 感觉是数据的问题，你的rosbag play [bag path]这个数据请问是怎么来的，采用mid70录制就好了嘛

Hi @Double-zh. It seems to be a data problem. The point cloud has no intensity. There is a rosbag under the v1.1 branch of this repo. You can play it to test it.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

xcc2731594 commented 5 days ago

I also want to know how...

do you know how

Livox-SDK / livox_detection

I want to add new class by own Livox data. How can I train? #21