Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Input nodes missing in frozen graph #114

Open
sjmjk429 opened this issue May 22, 2019 · 1 comment
Open

Input nodes missing in frozen graph #114

sjmjk429 opened this issue May 22, 2019 · 1 comment

Comments

@sjmjk429
Copy link

With TensorFlow v1.9.0, I went through the training instructions and produced the following .pbtxt and checkpoint files:

celoss.JPG
model.ckpt-104063.meta
model.ckpt-120000.data-00000-of-00002
checkpoint
model.ckpt-111901.data-00000-of-00002  
model.ckpt-120000.data-00001-of-00002
events.out.tfevents.1558022250.indy    
model.ckpt-111901.data-00001-of-00002  
model.ckpt-120000.index
graph.pbtxt                            
model.ckpt-111901.index                
model.ckpt-120000.meta
locloss.JPG                           
model.ckpt-111901.meta                 
model.ckpt-96213.data-00000-of-00002
loss.JPG                               
model.ckpt-119735.data-00000-of-00002  
model.ckpt-96213.data-00001-of-00002
model.ckpt-104063.data-00000-of-00002  
model.ckpt-119735.data-00001-of-00002 
model.ckpt-96213.index
model.ckpt-104063.data-00001-of-00002  
model.ckpt-119735.index                
model.ckpt-96213.meta
model.ckpt-104063.index                
model.ckpt-119735.meta                 
predict

I then tried to freeze the graph:

$ freeze_graph \
    --input_graph=./logs/graph.pbtxt \
    --input_binary=false \
    --input_checkpoint=./logs/model.ckpt-120000 \
    --output_graph=frozen_graph.pb \
    --output_node_names=ssd300/Reshape_13,ssd300/Reshape_12

2019-05-22 13:25:28.443832: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1392] Found device 0 with properties: 
name: GeForce GTX 1080 major: 6 minor: 1 memoryClockRate(GHz): 1.8475
pciBusID: 0000:07:00.0
totalMemory: 7.92GiB freeMemory: 7.35GiB
2019-05-22 13:25:28.443862: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1471] Adding visible gpu devices: 0
2019-05-22 13:25:28.766253: I tensorflow/core/common_runtime/gpu/gpu_device.cc:952] Device interconnect StreamExecutor with strength 1 edge matrix:
2019-05-22 13:25:28.766284: I tensorflow/core/common_runtime/gpu/gpu_device.cc:958]      0 
2019-05-22 13:25:28.766292: I tensorflow/core/common_runtime/gpu/gpu_device.cc:971] 0:   N 
2019-05-22 13:25:28.766383: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1084] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 7091 MB memory) -> physical GPU (device: 0, name: GeForce GTX 1080, pci bus id: 0000:07:00.0, compute capability: 6.1)

This seemed to successfully produce a frozen graph file:
105203994 May 21 13:55 frozen_graph.pb

When I inspected the frozen_graph.pb file, it appears that input nodes are missing from the graph:

$ decent_q inspect --input_frozen_graph frozen_graph.pb 

Op types used: 167 Const, 71 Identity, 35 BiasAdd, 35 Conv2D, 23 Relu, 15 Reshape, 12 Transpose, 12 StridedSlice, 12 Pack, 5 MaxPool, 2 Mul, 2 ConcatV2, 1 Maximum, 1 QueueDequeueManyV2, 1 FIFOQueueV2, 1 Rsqrt, 1 SpaceToBatchND, 1 Square, 1 Sum, 1 BatchToSpaceND
 
No inputs found.
Found 2 possible outputs: (name=ssd300/Reshape_12, op=Reshape) (name=ssd300/Reshape_13, op=Reshape) 

Any ideas as to why the input node(s) are missing? What should the input nodes for this model be?

Thank you.

@jamiechoi1995
Copy link

where did you feed the input image?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants