自贡网站推广,万户网站建设,网站建设实训个人,骏驰网站建设我是从keras入门深度学习的#xff0c;第一个用的demo是keras实现的yolov3#xff0c;代码很好懂#xff08;其实也不是很好懂#xff0c;第一次也搞了很久才弄懂#xff09; 然后是做的车牌识别#xff0c;用了tiny-yolo来检测车牌位置#xff0c;当时训练有4w张图片第一个用的demo是keras实现的yolov3代码很好懂其实也不是很好懂第一次也搞了很久才弄懂 然后是做的车牌识别用了tiny-yolo来检测车牌位置当时训练有4w张图片用了一天来训练当时觉得时间可能就是这么长也不懂GPU训练的时候GPU利用率所以不怎么在意后来随着项目图片片的增多训练时间越来越大受不了了看了一片文章才注意到GPU利用率的问题.想到要用tensorflow原生的api去训练比如用tf.data.dataset就找到了这个tensorflow原生实现yolo的项目在训练的时候发现他没加梯度衰减训练了一段时间total loss下不去了所以加了一个梯度衰减。想写一下文章小白的第一篇文章哈哈哈大神别喷我的内容太简单YunYang1994/tensorflow-yolov3github.com他好像改了train.py 原来是这样的import tensorflow as tf
from core import utils, yolov3
from core.dataset import dataset, Parser
sess tf.Session()IMAGE_H, IMAGE_W 416, 416
BATCH_SIZE 8
EPOCHS 2000*1000
LR 0.0001
SHUFFLE_SIZE 1000
CLASSES utils.read_coco_names(./data/voc.names)
ANCHORS utils.get_anchors(./data/voc_anchors.txt)
NUM_CLASSES len(CLASSES)train_tfrecord ../VOC/train/voc_train*.tfrecords
test_tfrecord ../VOC/test/voc_test*.tfrecordsparser Parser(IMAGE_H, IMAGE_W, ANCHORS, NUM_CLASSES)
trainset dataset(parser, train_tfrecord, BATCH_SIZE, shuffleSHUFFLE_SIZE)
testset dataset(parser, test_tfrecord , BATCH_SIZE, shuffleNone)is_training tf.placeholder(tf.bool)
example tf.cond(is_training, lambda: trainset.get_next(), lambda: testset.get_next())images, *y_true example
model yolov3.yolov3(NUM_CLASSES, ANCHORS)with tf.variable_scope(yolov3):y_pred model.forward(images, is_trainingis_training)loss model.compute_loss(y_pred, y_true)optimizer tf.train.AdamOptimizer(LR)
saver tf.train.Saver(max_to_keep2)tf.summary.scalar(loss/coord_loss, loss[1])
tf.summary.scalar(loss/sizes_loss, loss[2])
tf.summary.scalar(loss/confs_loss, loss[3])
tf.summary.scalar(loss/class_loss, loss[4])write_op tf.summary.merge_all()
writer_train tf.summary.FileWriter(./data/train)
writer_test tf.summary.FileWriter(./data/test)update_var tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, scopeyolov3/yolo-v3)
with tf.control_dependencies(update_var):train_op optimizer.minimize(loss[0], var_listupdate_var,global_stepglobal_step) # only update yolo layersess.run(tf.global_variables_initializer())
pretrained_weights tf.global_variables(scopeyolov3/darknet-53)
load_op utils.load_weights(var_listpretrained_weights,weights_file./darknet53.conv.74)
sess.run(load_op)for epoch in range(EPOCHS):run_items sess.run([train_op, write_op] loss, feed_dict{is_training:True})writer_train.add_summary(run_items[1], global_stepepoch)writer_train.flush() # Flushes the event file to diskif (epoch1)%1000 0: saver.save(sess, save_path./checkpoint/yolov3.ckpt, global_stepepoch)run_items sess.run([write_op] loss, feed_dict{is_training:False})writer_test.add_summary(run_items[0], global_stepepoch)writer_test.flush() # Flushes the event file to diskprint(EPOCH:%7d tloss_xy:%7.4f tloss_wh:%7.4f tloss_conf:%7.4f tloss_class:%7.4f%(epoch, run_items[2], run_items[3], run_items[4], run_items[5]))
然后我发现没有梯度下降所以就找了怎么实现实现如下
optimizer tf.train.AdamOptimizer(LR)
改为
global_step tf.Variable(0, trainableFalse)
learning_rate tf.train.exponential_decay(LR,100,0.93,staircaseTrue,global_stepglobal_step)
optimizer tf.train.AdamOptimizer(learning_rate)learningrate 是梯度的类LR是初始梯度100是每一百次初始梯度乘以衰减度这里是第三个参数0.93代表了衰减度globalstep_step global_step是一定要加的不然梯度一直保持了初始梯度。最后加个打印tf.summary.scalar(learning_rate,learning_rate)就可以爽快的去训练了
