Fu-En Wang

yes, the minimum loss is -1 so it is correct.

If you want to use your own dataset, I think the easiest way try to merge them into the same format as KITTI. You can download KITTI first and take...

Yes, you should run some object detection method like faster RCNN to extract the object 2D box. For each cropped box, you should resize them to 224*224 and then apply...

If you want to project 3D point to 2D image, you will need camera intrisic which is provided by KITTI. But I didn't include the infomation in my code. The...

1. According to the paper, there are two prediction from our model. The first is the revised angle for each bin of the circle (orient), the second one is the...

According to the KITTI documentation [here](http://www.cvlibs.net/publications/Geiger2013IJRR.pdf), The P2 of files in calib/ is the projection matrix of left color camera, which dimension is 3x4. This projection matrix is K[R|T] and...

1. For the confidence our model predict is try to know the confidence which bin the theta is actually located at. So the dimension will be [batch_size x num of...

@tingZ123 I just saw this issue now haha. In my eval code, I only eval the angle error and dimension error. If you want to obtain the 3D box, you...

For Q1,2,3, that is all correct. For Q4: You can print the Variable "batch"(https://github.com/fuenwang/3D-BoundingBox/blob/master/Eval.py#L56), the dimension will be [bs, ch, 224, 224] just change it to the image you want...

1. Yes, that is the same meaning. 2. chane the batch input image (https://github.com/fuenwang/3D-BoundingBox/blob/master/Eval.py#L56) and then the theta of this line is the global angle (https://github.com/fuenwang/3D-BoundingBox/blob/master/Eval.py#L76)