Comments (6)
@ardeal
We use caffe -time
to test the speed of FaceBoxes on CPU, which does not contain the time of pre-process (e.g., read/resize/.../ image) and post-process (e.g., NMS). Different implementations of NMS have differenet times. In actual use, you can set a very high score threshold (e.g., 0.8) to speed up NMS, or you can implement a faster NMS.
from faceboxes.
Many thanks for your reply. Dr. Zhang!
Do you have a trained GPU model?
from faceboxes.
@ardeal
The provided model is trained on GPU. You can test it on CPU or GPU.
from faceboxes.
Hi Dr. Zhang,
if I use the following python code:
caffe.set_mode_cpu()
the output of FaceBoxes algorithms is correct.
however, if I use the following code in python:
caffe.set_mode_gpu()
I got the following error while I initialize the network:
......
I0408 16:21:01.161911 31476 net.cpp:165] Memory required for data: 52188160
I0408 16:21:01.161911 31476 layer_factory.hpp:77] Creating layer mbox_conf_softmax
I0408 16:21:01.161911 31476 net.cpp:100] Creating Layer mbox_conf_softmax
I0408 16:21:01.161911 31476 net.cpp:434] mbox_conf_softmax <- mbox_conf_reshape
I0408 16:21:01.161911 31476 net.cpp:408] mbox_conf_softmax -> mbox_conf_softmax
I0408 16:21:01.161911 31476 net.cpp:150] Setting up mbox_conf_softmax
I0408 16:21:01.161911 31476 net.cpp:157] Top shape: 1 6400 2 (12800)
I0408 16:21:01.161911 31476 net.cpp:165] Memory required for data: 52239360
I0408 16:21:01.161911 31476 layer_factory.hpp:77] Creating layer mbox_conf_flatten
I0408 16:21:01.161911 31476 net.cpp:100] Creating Layer mbox_conf_flatten
I0408 16:21:01.161911 31476 net.cpp:434] mbox_conf_flatten <- mbox_conf_softmax
I0408 16:21:01.161911 31476 net.cpp:408] mbox_conf_flatten -> mbox_conf_flatten
I0408 16:21:01.161911 31476 net.cpp:150] Setting up mbox_conf_flatten
I0408 16:21:01.161911 31476 net.cpp:157] Top shape: 1 12800 (12800)
I0408 16:21:01.161911 31476 net.cpp:165] Memory required for data: 52290560
I0408 16:21:01.161911 31476 layer_factory.hpp:77] Creating layer detection_out
I0408 16:21:01.161911 31476 net.cpp:100] Creating Layer detection_out
I0408 16:21:01.161911 31476 net.cpp:434] detection_out <- mbox_loc
I0408 16:21:01.161911 31476 net.cpp:434] detection_out <- mbox_conf_flatten
I0408 16:21:01.161911 31476 net.cpp:434] detection_out <- mbox_priorbox
I0408 16:21:01.161911 31476 net.cpp:408] detection_out -> detection_out
I0408 16:21:01.161911 31476 net.cpp:150] Setting up detection_out
I0408 16:21:01.161911 31476 net.cpp:157] Top shape: 1 1 1 7 (7)
I0408 16:21:01.161911 31476 net.cpp:165] Memory required for data: 52290588
I0408 16:21:01.161911 31476 net.cpp:228] detection_out does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] mbox_conf_flatten does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] mbox_conf_softmax does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] mbox_conf_reshape does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] mbox_priorbox does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] mbox_conf does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] mbox_loc does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] conv4_2_mbox_priorbox does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] conv4_2_mbox_conf_flat does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] conv4_2_mbox_conf_perm does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] conv4_2_mbox_conf does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] conv4_2_mbox_loc_flat does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] conv4_2_mbox_loc_perm does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] conv4_2_mbox_loc does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] conv3_2_mbox_priorbox does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] conv3_2_mbox_conf_flat does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] conv3_2_mbox_conf_perm does not need backward computation.
I0408 16:21:01.161911 31476 net.cpp:228] conv3_2_mbox_conf does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv3_2_mbox_loc_flat does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv3_2_mbox_loc_perm does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv3_2_mbox_loc does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_concat_mbox_priorbox does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_concat_mbox_conf_flat does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_concat_mbox_conf_perm does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_concat_mbox_conf does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_concat_mbox_loc_flat does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_concat_mbox_loc_perm does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_concat_mbox_loc does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv4_2_conv4_2_relu_0_split does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv4_2_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv4_2_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv4_2_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv4_2 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv4_1_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv4_1_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv4_1_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv4_1 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv3_2_conv3_2_relu_0_split does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv3_2_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv3_2_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv3_2_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv3_2 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv3_1_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv3_1_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv3_1_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv3_1 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_concat_inception_a3_concat_0_split does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_concat does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_3_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_3_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_3_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_3 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_2_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_2_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_2_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_2 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_2_reduce_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_2_reduce_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_2_reduce_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_2_reduce does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_reduce_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_reduce_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_reduce_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_3x3_reduce does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_1x1_2_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_1x1_2_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_1x1_2_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_1x1_2 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_1x1_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_1x1_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_1x1_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_1x1 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a3_pool_ave does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_concat_inception_a2_concat_0_split does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_concat does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_3_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_3_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_3_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_3 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_2_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_2_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_2_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_2 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_2_reduce_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_2_reduce_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_2_reduce_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_2_reduce does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_reduce_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_reduce_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_reduce_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_3x3_reduce does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_1x1_2_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_1x1_2_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_1x1_2_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_1x1_2 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_1x1_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_1x1_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_1x1_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_1x1 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a2_pool_ave does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_concat_inception_a1_concat_0_split does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_concat does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_3_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_3_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_3_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_3 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_2_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_2_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_2_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_2 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_2_reduce_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_2_reduce_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_2_reduce_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_2_reduce does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_reduce_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_reduce_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_reduce_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_3x3_reduce does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_1x1_2_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_1x1_2_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_1x1_2_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_1x1_2 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_1x1_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_1x1_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_1x1_bn does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_1x1 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] inception_a1_pool_ave does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] pool2_pool2_0_split does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] pool2 does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv2_relu does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv2_scale does not need backward computation.
I0408 16:21:01.177520 31476 net.cpp:228] conv2_concat does not need backward computation.
I0408 16:21:01.205895 31476 net.cpp:228] conv2_neg does not need backward computation.
I0408 16:21:01.205895 31476 net.cpp:228] conv2_conv2_bn_0_split does not need backward computation.
I0408 16:21:01.205895 31476 net.cpp:228] conv2_bn does not need backward computation.
I0408 16:21:01.205895 31476 net.cpp:228] conv2 does not need backward computation.
I0408 16:21:01.205895 31476 net.cpp:228] pool1 does not need backward computation.
I0408 16:21:01.205895 31476 net.cpp:228] conv1_relu does not need backward computation.
I0408 16:21:01.205895 31476 net.cpp:228] conv1_scale does not need backward computation.
I0408 16:21:01.205895 31476 net.cpp:228] conv1_concat does not need backward computation.
I0408 16:21:01.205895 31476 net.cpp:228] conv1_neg does not need backward computation.
I0408 16:21:01.205895 31476 net.cpp:228] conv1_conv1_bn_0_split does not need backward computation.
I0408 16:21:01.205895 31476 net.cpp:228] conv1_bn does not need backward computation.
I0408 16:21:01.205895 31476 net.cpp:228] conv1 does not need backward computation.
I0408 16:21:01.205895 31476 net.cpp:228] data_input_0_split does not need backward computation.
I0408 16:21:01.205895 31476 net.cpp:228] input does not need backward computation.
I0408 16:21:01.205895 31476 net.cpp:270] This network produces output detection_out
I0408 16:21:01.205895 31476 net.cpp:283] Network initialization done.
I0408 16:21:01.221505 31476 upgrade_proto.cpp:77] Attempting to upgrade batch norm layers using deprecated params: ../models/faceboxes/faceboxes.caffemodel
I0408 16:21:01.221505 31476 upgrade_proto.cpp:80] Successfully upgraded batch norm layers using deprecated params.
I0408 16:21:01.221505 31476 net.cpp:761] Ignoring source layer data
I0408 16:21:01.221505 31476 net.cpp:761] Ignoring source layer data_data_0_split
I0408 16:21:01.221505 31476 net.cpp:761] Ignoring source layer mbox_loss
C:\ProgramData\Anaconda3\lib\site-packages\skimage\io\_io.py:49: UserWarning: `as_grey` has been deprecated in favor of `as_gray`
warn('`as_grey` has been deprecated in favor of `as_gray`')
F0408 16:21:01.371168 31476 im2col.cu:61] Check failed: error == cudaSuccess (48 vs. 0) no kernel image is available for execution on the device
*** Check failure stack trace: ***
Do you have any idea about the issue?
Thanks and Best Regards,
Ardeal
from faceboxes.
Hi Dr. Zhang,
I found out the reason of the issue:
in build.cmd file, we should specify the arch of GPU by:
if NOT DEFINED CUDA_ARCH_NAME set CUDA_ARCH_NAME=Pascal
CUDA_ARCH_NAME should be set to specified name according the GPU arch.
now, I could run GPU version of FaceBoes.
Many thanks!
from faceboxes.
@ardeal
OK
from faceboxes.
Related Issues (20)
- The algorithm in NVIDIA GTX1080TI is 25fps for 1280*120 pixels' images. HOT 1
- compile HOT 1
- 1378:17: Message type "caffe.LayerParameter" has no field named "permute_param" HOT 2
- Range of the output HOT 1
- about train my data
- About train on other data HOT 1
- Error 1 google/protobuf/port_def HOT 1
- 感受野的计算是否存在问题 HOT 2
- F0731 17:32:46.966609 19825 detection_output_layer.cpp:164] Check failed: num_priors_ * num_loc_classes_ * 4 == bottom[0]->channels() (4000 vs. 25600) Number of priors must match number of location predictions. HOT 1
- caffe version is the same as pytorch version HOT 2
- can we run this on raspberry pi and jetson nano kind of devices HOT 3
- FDDB without upsampling HOT 2
- windows 跑出来的结果不正确。 HOT 3
- migrate to tensorflow 2.0?
- Expand image augmentation HOT 2
- Can not download trained model HOT 1
- when release the model or the training with 5 landmarks?
- Mistakes in Readme
- Issue with prediction using OpenCV dnn module HOT 1
- use on pascal voc classes
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from faceboxes.