The actual number of training samples（部分.txt中为空，所以实际训练样本只有3万多）

[Lg955]

After split，there are 69337 imgs in trainval folder, but there are more than 30 thousand *.txt is empty. so when generating the trainval.txt, we can ignore the empty txt, then the useful train samples are 35462.

May be this is an effective method to reduce negative samples.

---

[yijingru]

The negative images have been involved in training. I guess you mean the label is empty?

[Lg955]

The negative images have been involved in training. I guess you mean the label is empty?

yeap, so can we ignore the imgs whose labels are empty?