tensorflow-yolov3 icon indicating copy to clipboard operation
tensorflow-yolov3 copied to clipboard

🔥 TensorFlow Code for technical report: "YOLOv3: An Incremental Improvement"

Results 114 tensorflow-yolov3 issues
Sort by recently updated
recently updated
newest added

我尝试用tensorrt对yolov3进行加速,利用tensorflow自带的api将yolov3_coco.pb转为为yolov3_trt.pb,但是出现一个问题,程序无法识别模型的尺寸,会出现[?,?,?,3,2]这种情况。 ![image](https://user-images.githubusercontent.com/39334643/61201157-39a8e580-a716-11e9-9fd0-9f4cbbbe863c.png) 然后我查了一下网络结构(最下图是能转化的模型),发现缺少一些尺寸信息,请问要如何获取这些type信息呢?感谢大佬了 ![image](https://user-images.githubusercontent.com/39334643/61202382-f2245880-a719-11e9-837e-b4b6960f4ab7.png) ![image](https://user-images.githubusercontent.com/39334643/61202393-fc465700-a719-11e9-8501-4c9613d0a105.png)

请问,我训练的类别是3类,原数据量是1573,做了四个方向的数据增强后,数据量为六千多,但是训练到后面,train loss和test loss基本趋向于稳定,大概是12-15,有什么方法把loss下降,是数据量不够还是为什么? ![image](https://user-images.githubusercontent.com/32948294/85991451-04c89800-ba26-11ea-9acc-90cac51313c1.png)

训练了30个epoch,mAP能达到0.72 ![mAP](https://user-images.githubusercontent.com/33013904/58149350-2be65f00-7c95-11e9-9d07-b4612df5526a.png)

根据focalloss的原理 应该写成如下形式 focal_loss = tf.abs(target - alpha) * tf.pow(tf.abs(target - actual), gamma) # 正样本时为 (1-alpha) * tf.pow(actual, gamma) # 负样本时为 (alpha) * tf.pow((1-actual), gamma) 假如按照你的写法 focal_loss = alpha * tf.pow(tf.abs(target...

Hi, Your work is excellent! I have seen your result of VOC2012, mAP=88.38% according to the [blog](https://yunyang1994.github.io/posts/YOLOv3/#more). But I cannot find your result on official VOC2012 [leaderboard](http://host.robots.ox.ac.uk:8080/leaderboard/displaylb_main.php?challengeid=11&compid=3)

请问有人有VOC数据集的预训练权重吗?能否分享一份 QQ 1552006123

I am clearly observing decrease in train loss and test loss while training but when I'm running evaluate.py it just prints ground truth , i have put some print statements...

训练自己的数据集的时候,生成train.txt的时候的class_id从0开始还是从1开始?比如我有30个类别的图片,那么做这个train.txt的时候class_id从几开始?

I have trained yolov3 for face detection with WIDER face dataset. I want to detect small faces. When I trained with pretrained weight from coco dataset the loss was too...

I am testing on VOC2007 dataset. The training went ok and the training loss is keep dropping in a good way (after 3 epoch it was ~30). However, every time...