xiaowenhe
xiaowenhe
when I change Resize (1600,900) to (1200,675) in train_pipeline = [ dict(type='LoadImageFromFileMono3D'), dict( type='LoadAnnotations3D', with_bbox=True, with_label=True, with_attr_label=True, with_bbox_3d=True, with_label_3d=True, with_bbox_depth=True), #dict(type='Resize', img_scale=(1600, 900), keep_ratio=True), dict(type='Resize', img_scale=(1200, 675), keep_ratio=True), dict(type='RandomFlip3D', flip_ratio_bev_horizontal=0.5),...
Does this project only support SSD_mobilenet and ssd_inception_v2 when detection? How to use faster-rcnn? Thank you!
Hi,@sergiomsilva, if one car has two license plate, how can i train the wpod-net? like the following pictures: 
hello ,@sergiomsilva, could you release the OCR training code now? I follow your WPOD-NET, and get good results in our datasets, but this is only LP detection. And OCR result...
你好， 非常感谢你的工作。上周用你的代码训练了其他任务的检测和关键点定位模型，但是在训练的过程中，总是出现 《读取图片的过程中，线程一直等待，无法继续训练问题》， 图片内容可能不能显示，具体内容如下： 第一次训练： 2244: 2.664922，2.490456 avg loss, 0.001000 rate, 0.701170 seconds, 71808 images, 62.321252 hours left 第二次训练： 2256: 2.414187， 2.562203 avg loss, 0.001000 rate, 0.712135 seconds, 72192 images,...
你好，感谢你的开源。我在阅读你的代码的过程中，对anchor和gt的匹配存在一点疑问，麻烦你能帮忙解释下吗？ 在anchor.py的231行，sorted_ious = sorted_ious[np.logical_not(anchor_matched_already)]这句代码的意义是什么？好像并没有什么关联与box_ious?
在dataset.py 第234行计算IOU的时候 iou_scale = self.bbox_iou(bbox_xywh_scaled[i][np.newaxis, :], anchors_xywh)，代码中用的anchor是实际的相对与实际坐标的anchor大小，比如【10,13, 16,30, 33,23, 30,61, 62,45, 59,119, 116,90, 156,198, 373,326】，而bbox_xywh_scaled确是除以strides之后的，这样的话，是否存在矛盾呢？ 比如，某gt的宽高都为608，训练时的图片最大宽高，假设gt占满全图。则608/8 = 76。也就是说，bbox_xywh_scaled里面的宽高最大才78，那anchor里面的 大于这个宽高的anchor还有什么用呢？所以，这个地方计算iou的时候感觉存在问题。 以上是个人理解，有不对的，请谅解！
Thank you!
Hi, @authors, Thanks for your works. Now I want to use my own datasets to train. So first I maybe to do my datasets as ILSVRC2015. But I cannot download...
Hi,I have a question, what is the difference as follows: #implemented by py_func #value = tf.identity(xw) #substract the marigin and scale it value = coco_func(xw_norm,y,alpha) * scale #implemented by tf...