Closed RuikunLi closed 3 years ago
There are many same requests, so when the time comes, I will write and upload a guideline to train only the classes I want in Coco.
train2017.zip (images) http://images.cocodataset.org/zips/train2017.zip val2017.zip (images) http://images.cocodataset.org/zips/val2017.zip instances_train2017.json, instances_val2017.json http://images.cocodataset.org/annotations/annotations_trainval2017.zip
Ref: https://wiki.loliot.net/docs/lang/python/libraries/yolov4/python-yolov4-dataset
Copy the conversion script in the above link.
INSTANCES_PATH = "instances_train2017.json"
NAMES_PATH = "custom.names"
OUTPUT_FILE_PATH = "custom_train2017.txt"
After editing the above part of the script, run the script for train
and val
respectively.
Then, you can get custom_train2017.txt
and custom_val2017.txt
. It will probably take an hour or two.
Ref: https://wiki.loliot.net/docs/lang/python/libraries/yolov4/python-yolov4-training
Set epochs to the number of classes * 5
or more.
Hi, I would like to know if it is possible to detect customer class id.
For example, I only interest in the person in an image, is it possible to only detect the person and draw the person bboxes?
thx in advance