exiawsh/StreamPETR
[ICCV 2023] StreamPETR: Exploring Object-Centric Temporal Modeling for Efficient Multi-View 3D Object Detection
PythonNOASSERTION
Issues
- 2
推理时是不是没有使用历史帧?
#220 opened by MargeryLab - 1
为什么不使用GIoULoss
#219 opened by SmallXieGithub - 4
StreamPetr Tracking
#218 opened by Gaondong - 1
- 4
ImportError: libtorch_cuda_cu.so: cannot open shared object file: No such file or directory
#213 opened by Vaishnav-B - 4
请问如何根据输入的图像size和最终的size来设置resize_lim
#215 opened by SmallXieGithub - 1
关于Focal-PETR中的注意力图
#216 opened by flyingdrone - 2
- 6
train+val作为训练集时,多卡训练NCCL超时报错,大佬可以看看吗?
#212 opened by loserhou - 0
可以提供使用vov-99训练在测试集上的log吗
#211 opened by loserhou - 2
在测试集上V2-99那个版本训练了60个epoch吗?希望得到您的解答
#210 opened by loserhou - 1
关于map=55.0版本,请问是使用train➕val作为训练集吗?
#209 opened by loserhou - 1
RepDETR3D连续帧采样逻辑
#208 opened by sms95 - 3
temporal_alignment里当前query_pos应用MLN的目的是啥?
#205 opened by xuanzhangyang - 1
streampetr_head.py中cone的作用是什麼?
#206 opened by jinpeifei2015 - 2
2.5D center
#207 opened by 1yuechu1 - 2
projects/configs/StreamPETR/stream_petr_vov_flash_800_bs2_seq_24e.py文件里用的是MultiHeadAttention,但是projects/mmdet3d_plugin/models/utils/petr_transformer.py里面没有MultiHeadAttention,只有PETRMultiheadAttention
#204 opened by chenxinhe1 - 9
projects/configs/StreamPETR/stream_petr_vov_flash_800_bs2_seq_24e.py文件里用的是MultiHeadAttention,但是projects/mmdet3d_plugin/models/utils/petr_transformer.py里面没有MultiHeadAttention,只有PETRMultiheadAttention
#203 opened by chenxinhe1 - 1
Running tracking evaluation
#194 opened by ekosman - 0
projects/configs/StreamPETR/stream_petr_vov_flash_800_bs2_seq_24e.py文件里用的MultiHeadAttention,但是projects/mmdet3d_plugin/models/utils/petr_transformer.py里面没有MultiHeadAttention,只有PETRMultiheadAttention
#202 opened by chenxinhe1 - 2
About the commands using in the “docs/training_inference.md” and the training results
#201 opened by MillionaireChen - 1
关于focalhead
#197 opened by loserhou - 0
About test on nuscenes
#200 opened by Jason-Du111 - 6
- 0
nuscenes v1.0mini is ok. But "No such file or directory: './data/nuscenes/v1.0-test/visibility.json'" appears when i ran v1.0.
#198 opened by MillionaireChen - 2
感谢您的优秀工作,关于reset memory我有个问题,通过调试我发现是窗口20帧,但每个batch都会进行reset重置为none,请问memory bank是怎么更新的呢?
#195 opened by loserhou - 1
关于 flash attention版本导致 VOVNet 为backbone的收敛问题
#196 opened by nancy618618 - 1
- 1
What is the meaning of config file?
#192 opened by Jason-Du111 - 6
- 1
Validity of using gray images :)
#191 opened by sjminInSwm - 7
远处目标有较多重影,近处同向目标较为正常,应如何解决
#186 opened by Gaondong - 4
- 2
[问题] gt_bboxes_3d是如何由sample_annotation.json转换得到的?
#188 opened by LiuJieShane - 8
How to calculate the 'resize_lim' paras?
#183 opened by cmiforsaken - 2
Precision and recall for all the classes
#187 opened by GUE1RNG - 2
training errors and settings
#185 opened by Terencedu - 4
position_embeding question
#184 opened by xuanzhangyang - 1
- 3
R101 config
#173 opened by NaomiEX - 3
KeyError: 'Petr3D_seg: "PETRHead_seg: \'SinePositionalEncoding3D is not in the position encoding registry\'"'
#179 opened by chenxinhe1 - 3
about customdataset
#178 opened by XLiedInfi - 3
- 3
Input only the desired sequence
#176 opened by GUE1RNG - 4
- 1
- 8
- 2
Missing rate = 0.46, if I use IoU as the match rule
#177 opened by Terencedu - 4
- 4
RuntimeError: CUDA error: operation not supported when calling `cusparseCreate(handle)`
#172 opened by chenxinhe1