leon-liangwu / MaskYolo_Caffe

YOLO V2 & V3 , YOLO Combined with RCNN and MaskRCNN

114 stars 50 forks source link

path/to/train.txt #20

Closed marekjaszuk closed 4 years ago

marekjaszuk commented 4 years ago

Hi, I'm trying to train the object detector, but I got stuck on generating the lmdb through the command:

generate lmdb for detection

sh ./scripts/convert_detection.sh /path/to/train.txt /path/to/lmdb The problem is that I do not know what the /path/to/train.txt should poin to? I have the VOC data, and I was trying to point various locations and .txt files, but I still get the following message: convert_dec_data.cpp:86] A total of 0 images. So as it seems it does not find the proper file.

leon-liangwu commented 4 years ago

@marekjaszuk Firstly, you need to use voc_label.py to generate train.txt which is a list of the image paths. Every line of this train.txt is a path to an image. Then you replace /path/to/train.txt to the real path to your train.txt.