search

TimoSaemann / ENet

ENet: A Deep Neural Network Architecture for Real-Time Semantic Segmentation
583 stars 275 forks source link

caffe train -solver error: Check failed: bottom[i]->shape() == bottom[0]->shape() #19

Closed SergeMv closed 7 years ago

SergeMv commented 7 years ago

Hi Timo,

Got this when was running this command: /root/ENet/caffe-enet/build/tools/caffe train -solver /root/ENet/prototxts/enet_solver_encoder.prototxt

I0711 14:14:57.111349 209 net.cpp:100] Creating Layer eltwise2_0_4 I0711 14:14:57.111353 209 net.cpp:434] eltwise2_0_4 <- drop2_0_3 I0711 14:14:57.111371 209 net.cpp:434] eltwise2_0_4 <- bn2_0_4 I0711 14:14:57.111377 209 net.cpp:408] eltwise2_0_4 -> eltwise2_0_4 F0711 14:14:57.111387 209 eltwise_layer.cpp:34] Check failed: bottom[i]->shape() == bottom[0]->shape() Check failure stack trace: @ 0x7f6f05d78daa (unknown) @ 0x7f6f05d78ce4 (unknown) @ 0x7f6f05d786e6 (unknown) @ 0x7f6f05d7b687 (unknown) @ 0x7f6f0644fbbd caffe::EltwiseLayer<>::Reshape() @ 0x7f6f06317835 caffe::Net<>::Init() @ 0x7f6f063186c5 caffe::Net<>::Net() @ 0x7f6f063296ca caffe::Solver<>::InitTrainNet() @ 0x7f6f0632a8dc caffe::Solver<>::Init() @ 0x7f6f0632ac0a caffe::Solver<>::Solver() @ 0x7f6f062eab23 caffe::Creator_AdamSolver<>() @ 0x411fd6 caffe::SolverRegistry<>::CreateSolver() @ 0x40af52 train() @ 0x40898c main @ 0x7f6f0456ef45 (unknown) @ 0x409293 (unknown) @ (nil) (unknown) Aborted (core dumped)

What could be wrong?

TimoSaemann commented 7 years ago

I think the problem is your image size. The image size needs to be divisible three times by 2 without remainder. For example: 

input image size: 1024x512
after first pool: 512x256
after second pool: 256x128
after third pool: 128x64

after first upsample: 256x128
after second upsample: 512x256
after third upsample: 1024x512 (the same size as your input image)

If your input image has for example 500x220 :

after first pool: 250x110
after second pool: 125x65
after third pool: 63x33

after first upsample: 126x66
after second upsample: 252x132
after third upsample: 504x264 (not the same size as your input image)

Somewhere in the net you get this error: Check failed: bottom[i]->shape() == bottom[0]->shape()

In general it should be possible to use every image size, but you have to adapt the network with padding. For example you can expand the upsample_param with _upsamplew and _upsampleh (cf. https://github.com/alexgkendall/SegNet-Tutorial/issues/37).

SergeMv commented 7 years ago

Yes, I changed the image size and it looks like this step is working. Thank you.