Cascade_FPN_Tensorflow

Cascade_FPN_Tensorflow copied to clipboard

修改了cfgs文件中的eval_threshold与voc_eval文件里面的ovthresh参数都没有变化，谢谢！

homxxx

运行十次左右step后sess.run被挂住

2

你好！十分感谢你的代码分享。我先使用VOC2007进行训练一下看一下结果，但是在读取数据十次step后，sess.run后出现了代码不运行的状态，不报错也不跳出，请问一下具体可能会在哪个方面出了问题呢？我找了挺久的问题的。。。实在不知道怎么改。感谢！！！

yellowsmallblue

代码中设置的MAX_ITERATION = 20*SAVE_WEIGHTS_INTE，其中SAVE_WEIGHTS_INTE=80000.那就是说迭代次数设置了160万次，需要训练这么多次吗？我设置50万次，total _loss一直没有收敛，大概在0.02-0.1之间震荡

xinlanzuo

Error “Invalid argument: Incompatible shapes” when I try to run eval.py to evaluate trained weights.

1

I trained this network in my dataset, but when I try to evaluate the trained weights using eval.py, the error occured, can anyone help me? Thanks a lot!! The details...

Yansslc

How to save the model into pb model

2

Hi, I want to save the model into pb model, some information from Internet says I need to use the output node to save it, but it hard to find...

herrickli

can it used in DOTA dataset?

9

Thanks for your work：）and I wanna know can I use it in DOTA dataset?Cause I see the code structure is similar with r2cnn.If it can be used in DOTA,where should...

Smoothing97

train without pretrain weights。。。

1

I don't want to use the weight of the pre-training to train the model, where can I change； I modified ： # restorer, restore_ckpt = faster_rcnn.get_restorer() # if not restorer...

AiIsBetter

Training automatically closes

1

hello,author. i meet a question when i run your code(GTX 1080 Ti), sometime report this error 2019-12-12 14:41:58.024277: E tensorflow/stream_executor/cuda/cuda_dnn.cc:81] 2019-12-12 14:41:58.024505: I tensorflow/stream_executor/stream.cc:4818] stream 0x560dcbf2bd50 did not memzero GPU...

g18tka17

question about batch

3

thanks your project.I improved my batch size ,but there are nothing changes in my GPU used-memory,did it worked?what i can do? but i am sure that my tf was in...

zhudd-hub

Out of range error, when restore model

1

大佬您好，我之前跑了您的Faster Rcnn和Casecade的代码，现在尝试Casecade_FPN时遇到了Out of range的问题，我的数据集是voc格式的，也可以用convert_data_to_tfrecord.py和convert_data_to_tfrecord_voc2012.py进行数据转化，并且能够在tfrecord下生成文件。我看readme里您用的是coco数据集进行训练，请问我这个outofrange的报错是因为数据格式不匹配吗？因为这组数据在Faster和casecade上都可以训练，所以我排出了图片和label对应不准的问题。谢谢您！希望得到您的回复！

liujiaming1996