Yifan Zhang

自己创建外键，而不是intellij自动创建的BlogsById和userByUserId

> I changed the score threshold (0.78->0.6) and now it reaches 85 AP11 as below > > bbox AP:97.9524, 89.7096, 89.3596 bev AP:90.3292, 88.2457, 87.8269 3d AP:89.5458, 85.6965, 78.9677 aos...

> Thanks for sharing. @fregu856 You apply EBM-based regression to 3DOD by extending the SA-SSD with a conditional EBM p(y|x; θ). Why don't you just use a DNN f(θ) to...

I have another question: Have you ever tried to train the SA-SSD with the extended architecture simultaneously, rather than fine tune the EBM-related module on the fixed SA-SSD checkpoint.

@wei-tim Do you set USE_ROAD_PLANE=True, according to my experiments this parameter is crucial.

> Hi @wei-tim , > I have trained the model with the released code. > I can reproduce the results. > This is my log file: https://drive.google.com/file/d/1OE_K_m_ugztewieCsZ9zasZOZlsToW8M/view?usp=sharing. > Hope this...

I have the same error too when I use the MinkowskiEngine with tag v0.5.4. Then I checkout commit 02fc608bea4c0549b0a7b00ca1bf15dee4a0b228 and re-install the MinkowskiEngine, and the error disappears.

> Hello @nazMahmoud . The model that predicted 'cascade_rcnn_sigmoid_data' is trained on training (3712 examples) set. You are right, cascaded-rcnn repo doesn't provide that, so, I trained it by myself....

Hi @pangsu0613 , Thanks for your reply. And I have another question. Do you have plan to publish codebase to support other 3D detectors (PV-RCNN and CT3D)? Or do you...

Thanks for your suggestion. It is a typo. The mAP ratio is the ratio of mAP on adversarial examples to that on clean point cloud over the whole validation set....