codemaster

As in readme of this repo, the precision table is the weight converted from google automl's. But what is the precision if train on imagenet directly

Just my training result now

10

Training speed very slow, I have only one gpu titanx, have trained for three days. Train Epoch: 186 [128/960 (13%)] Loss: 1.369987 Detection Loss: 0.646443 Recognition Loss:0.723544 Train Epoch: 186...

Detection loss should not divide all element

4

In loss.py, detection loss is: return torch.mean(L_g * y_true_cls * training_mask) + classification_loss But I think the loss mean should divide only nonzero(y_true_cls * training_mask), so may be the loss...

Which version of pytorch should be used?

2

Hello, I run the code with some error as follows, so may I know your version of pytorch? `Memory Usage: CUDA: 0 Allocated: 117.5244140625 MB Cached: 125.875 MB Traceback (most...

After successfully compiled, the demo generate image with too many boxes

5


Very slow, use new version caffe2

5

1. I clone new caffe2, compile a cpu version, and have: libCaffe2_CPU.a libCAFFE2_NNPACK.a libCAFFE2_PTHREADPOOL.a libcpufeatures.a 2. create script.ar: CREATE libcaffe2.a ADDLIB libCAFFE2_NNPACK.a ADDLIB libCAFFE2_PTHREADPOOL.a ADDLIB libcpufeatures.a ADDLIB libCaffe2_CPU.a SAVE END...

只用两张图片训练，竟然不收敛，感觉不可能呢

3

作者提供的图片，只用两张训练是收敛了，可是用我准备的，两张车牌图片没有收敛，与作者原训练任务不同点就是车牌图片模糊些，id范围只有一个中文省份，和数字和大写字母，结果不能收敛 就算随机猜，也能50%,可是trainloss 1.0左右，acc为0，为什么呢？ `1/1 [==============================] - 0s 255ms/step - loss: 1.4361 - acc: 0.0000e+00 - val_loss: 1.4272 - val_acc: 0.0000e+00 Epoch 18485/100000000 ` 而训练集只有一张图片，train loss 可以到0.01， train acc 到1.0 ![ch11_20191230044803...

Hello, thanks for your effort. Could you provide tensorlfow 1.15.0 on rk3399? Thank you in advance

clsInfo应该把返回类别都存入吧

1

https://github.com/RaySue/NNIE-lite/blob/master/examples/GeneralCls.cpp#L111 您好，请问这句push_back, 是否应该放在if前面，存错所有类别得分，index里面保存最大值得索引。

NNIE_Get_Result 偶尔会输出远超4096的值

4

网络多数时候运行是正常的，偶尔会输出远超4096的值。 https://github.com/RaySue/NNIE-lite/blob/daf0dc19f47bc1e286308d137e1e636c2ef98da7/src/nnie_core.c#L272 ``` for (n = 0; n < pstNnieParam->astSegData[u32SegIdx].astDst[u32NodeIdx].u32Num; n++) { for (i = 0; i < u32Chn; i++) { for (j = 0; j < u32Height; j++) {...