-
Notifications
You must be signed in to change notification settings - Fork 51
Description
Hello,
I don't know what is wrong with the
https://github.com/Robert-JunWang/PeleeNet/blob/master/caffe/peleenet.caffemodel
file but I get a zero accuracy result when I run it in caffe using the ImageNet dataset.
I tried my setup with other models and I get valid results, so it doesn't seem to be a caffe setup issue.
Could you please confirm that there is nothing wrong with your caffemodel file or perform a forward pass to make sure that the problem is from my side?
Thanks in advance,
Panagiotis
P.S.: I show my caffe prototxt data layer and some of the inference results.
prototxt data layer:
layer {
name: "data"
type: "Data"
top: "data"
top: "label"
include {
phase: TEST
}
transform_param {
crop_size: 224
mean_value: 104
mean_value: 117
mean_value: 123
}
data_param {
source: "caffe/examples/imagenet/ilsvrc12_val_lmdb"
batch_size: 32 #not *iter_size
backend: LMDB
}
}
sample results:
0210 19:37:09.175479 23329 caffe.cpp:304] Batch 38, accuracy_top5 = 0.03125
I0210 19:37:09.234783 23329 caffe.cpp:304] Batch 39, accuracy_top1 = 0
I0210 19:37:09.234794 23329 caffe.cpp:304] Batch 39, accuracy_top5 = 0.03125
I0210 19:37:09.294126 23329 caffe.cpp:304] Batch 40, accuracy_top1 = 0
I0210 19:37:09.294137 23329 caffe.cpp:304] Batch 40, accuracy_top5 = 0.03125
I0210 19:37:09.353266 23329 caffe.cpp:304] Batch 41, accuracy_top1 = 0
I0210 19:37:09.353276 23329 caffe.cpp:304] Batch 41, accuracy_top5 = 0
I0210 19:37:09.412345 23329 caffe.cpp:304] Batch 42, accuracy_top1 = 0
I0210 19:37:09.412356 23329 caffe.cpp:304] Batch 42, accuracy_top5 = 0
I0210 19:37:09.471475 23329 caffe.cpp:304] Batch 43, accuracy_top1 = 0
I0210 19:37:09.471486 23329 caffe.cpp:304] Batch 43, accuracy_top5 = 0
I0210 19:37:09.530666 23329 caffe.cpp:304] Batch 44, accuracy_top1 = 0
I0210 19:37:09.530678 23329 caffe.cpp:304] Batch 44, accuracy_top5 = 0
I0210 19:37:09.589838 23329 caffe.cpp:304] Batch 45, accuracy_top1 = 0
I0210 19:37:09.589849 23329 caffe.cpp:304] Batch 45, accuracy_top5 = 0
I0210 19:37:09.649101 23329 caffe.cpp:304] Batch 46, accuracy_top1 = 0
I0210 19:37:09.649111 23329 caffe.cpp:304] Batch 46, accuracy_top5 = 0.03125
I0210 19:37:09.708271 23329 caffe.cpp:304] Batch 47, accuracy_top1 = 0
I0210 19:37:09.708281 23329 caffe.cpp:304] Batch 47, accuracy_top5 = 0