darknet_AlexeyAB_191221 - person - yolov3-tiny
1. anchors
2. yolov3-tiny.data
/home/yongqiang/darknet_work/darknet_AlexeyAB_191221/darknet-master/data/yolov3-tiny.data
classes= 1
train = /media/famu/DISK_DEEP/yongqiang/yolov3-tiny_prn/yolov3-tiny_prn.txt
# valid = /media/famu/DISK_DEEP/yongqiang/yolov3-tiny_prn/training_validation_test_sets_person_v0.txt
valid = /media/famu/DISK_DEEP/yongqiang/yolov3-tiny_prn/training_validation_test_sets_person_v0.txt
names = data/yolov3-tiny.names
backup = /media/famu/DISK_DATA/yongqiang/backup_yolov3-tiny
eval=coco
3. yolov3-tiny.names
/home/yongqiang/darknet_work/darknet_AlexeyAB_191221/darknet-master/data/yolov3-tiny.names
person
4. To calculate anchors
./darknet detector calc_anchors data/yolov3-tiny.data -num_of_clusters 8 -width 608 -height 608
(base) yongqiang@famu-sys:~/darknet_work/darknet_AlexeyAB_191221/darknet-master$ ./darknet detector calc_anchors data/yolov3-tiny.data -num_of_clusters 8 -width 608 -height 608
num_of_clusters = 8, width = 608, height = 608
read labels from 130656 images
loaded image: 130656 box: 1434742
all loaded.
calculating k-means++ ...
iterations = 133
avg IoU = 73.83 %
Saving anchors to the file: anchors.txt
anchors = 11, 42, 17, 75, 23,115, 33, 80, 35,155, 57,210, 119,347, 281,485
^C
(base) yongqiang@famu-sys:~/darknet_work/darknet_AlexeyAB_191221/darknet-master$
5. Training - ./darknet detector train ./train_cfg/yolov3-tiny.data ./train_cfg/yolov3-tiny.cfg -gpus 0,1,2,3 -map
Just train with -map flag:
darknet.exe detector train data/obj.data yolo-obj.cfg darknet53.conv.74 -map
So you will see mAP-chart (red-line) in the Loss-chart Window. mAP will be calculated for each 4 Epochs using valid=valid.txt file that is specified in obj.data file (1 Epoch = images_in_train_txt / batch iterations)
(to change the max x-axis value - change max_batches= parameter to 2000*classes, f.e. max_batches=6000 for 3 classes)
6. mAP
[net]
# Testing
# batch=1
# subdivisions=1
# Training
batch=64
subdivisions=16
width=608
height=608
channels=3
momentum=0.9
decay=0.0005
angle=0
saturation = 1.5
exposure = 1.5
hue=.1
learning_rate=0.001
burn_in=1000
max_batches = 500200
policy=steps
steps=400000,450000
scales=.1,.1
......
来源：CSDN
作者：ForeverStrong
链接：https://blog.csdn.net/chengyq116/article/details/103904471