Problem defining output tensor

[DanielJean007]

I've converted my darknet using: https://github.com/jinyu121/DW2TF. Which gives me the following files:

• yolov3-customv1.ckpt.data-00000-of-00001
• yolov3-customv1.ckpt.index
• yolov3-customv1.ckpt.meta
• yolov3-customv1.pb I then call your script with: with tf.Session(config=tf.ConfigProto(gpu_options=tf.GPUOptions(per_process_gpu_memory_fraction=0.50))) as sess: saver = tf.train.import_meta_graph("/home/nvidia/Downloads/DW2TF/data/yolov3-customv1.ckpt.meta") saver.restore(sess, "yolov3-customv1.ckpt") your_outputs = ["output_tensor/random"] your_outputs = ["output_tensor/Softmax"] frozen_graph = tf.graph_util.convert_variables_to_constants( sess, # session tf.get_default_graph().as_graph_def(),# graph+weight from the session output_node_names=your_outputs) with gfile.FastGFile("./model/frozen_model.pb", 'wb') as f: f.write(frozen_graph.SerializeToString()) print("Frozen model is successfully stored!")

Then, I receive the following error: AssertionError: output_tensor/Softmax is not in graph

So, I'm not sure where it needs to change. Could anyone help me, please?

My Yolov3.cfg can be found at: https://github.com/pjreddie/darknet/blob/master/cfg/yolov3.cfg

Thanks.

[ardianumam]

In object detection, we need at least 3 tensors to be called: (i) input, (ii) class prediction and (iii) box prediction. In this case, we need to know those tensor names when calling them. How to know those names? It depends who makes (who gives the tensor name). This repo uses model from here. I can know the tensor name outputs by seeing how this repo call frozen model to do prediction.

[DanielJean007]

Hi there. Thanks for the quick response.

Unfortunately, I still don't get where I can take those names from. Follow my .cfg file:

[net] # Testing batch=1 subdivisions=1 # Training # batch=64 # subdivisions=16 width=286 height=286 channels=3 momentum=0.9 decay=0.0005 angle=0 saturation = 1.5 exposure = 1.5 hue=.1

learning_rate=0.001 burn_in=1000 max_batches = 500200 policy=steps steps=400000,450000 scales=.1,.1

[convolutional] batch_normalize=1 filters=32 size=3 stride=1 pad=1 activation=leaky

# Downsample

[convolutional] batch_normalize=1 filters=64 size=3 stride=2 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=32 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=64 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

# Downsample

[convolutional] batch_normalize=1 filters=128 size=3 stride=2 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=64 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=128 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=64 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=128 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

# Downsample

[convolutional] batch_normalize=1 filters=256 size=3 stride=2 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=128 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=256 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=128 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=256 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=128 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=256 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=128 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=256 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=128 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=256 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=128 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=256 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=128 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=256 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=128 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=256 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

# Downsample

[convolutional] batch_normalize=1 filters=512 size=3 stride=2 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=256 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=512 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=256 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=512 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=256 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=512 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=256 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=512 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=256 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=512 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=256 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=512 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=256 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=512 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=256 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=512 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

# Downsample

[convolutional] batch_normalize=1 filters=1024 size=3 stride=2 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=512 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=1024 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=512 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=1024 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=512 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=1024 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

[convolutional] batch_normalize=1 filters=512 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 filters=1024 size=3 stride=1 pad=1 activation=leaky

[shortcut] from=-3 activation=linear

######################

[convolutional] batch_normalize=1 filters=512 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 size=3 stride=1 pad=1 filters=1024 activation=leaky

[convolutional] batch_normalize=1 filters=512 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 size=3 stride=1 pad=1 filters=1024 activation=leaky

[convolutional] batch_normalize=1 filters=512 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 size=3 stride=1 pad=1 filters=1024 activation=leaky

[convolutional] size=1 stride=1 pad=1 filters=255 activation=linear

[yolo] mask = 6,7,8 anchors = 10,13, 16,30, 33,23, 30,61, 62,45, 59,119, 116,90, 156,198, 373,326 classes=80 num=9 jitter=.3 ignore_thresh = .7 truth_thresh = 1 random=1

[route] layers = -4

[convolutional] batch_normalize=1 filters=256 size=1 stride=1 pad=1 activation=leaky

[upsample] stride=2

[route] layers = -1, 61

[convolutional] batch_normalize=1 filters=256 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 size=3 stride=1 pad=1 filters=512 activation=leaky

[convolutional] batch_normalize=1 filters=256 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 size=3 stride=1 pad=1 filters=512 activation=leaky

[convolutional] batch_normalize=1 filters=256 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 size=3 stride=1 pad=1 filters=512 activation=leaky

[convolutional] size=1 stride=1 pad=1 filters=255 activation=linear

[yolo] mask = 3,4,5 anchors = 10,13, 16,30, 33,23, 30,61, 62,45, 59,119, 116,90, 156,198, 373,326 classes=80 num=9 jitter=.3 ignore_thresh = .7 truth_thresh = 1 random=1

[route] layers = -4

[convolutional] batch_normalize=1 filters=128 size=1 stride=1 pad=1 activation=leaky

[upsample] stride=2

[route] layers = -1, 36

[convolutional] batch_normalize=1 filters=128 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 size=3 stride=1 pad=1 filters=256 activation=leaky

[convolutional] batch_normalize=1 filters=128 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 size=3 stride=1 pad=1 filters=256 activation=leaky

[convolutional] batch_normalize=1 filters=128 size=1 stride=1 pad=1 activation=leaky

[convolutional] batch_normalize=1 size=3 stride=1 pad=1 filters=256 activation=leaky

[convolutional] size=1 stride=1 pad=1 filters=255 activation=linear

[yolo] mask = 0,1,2 anchors = 10,13, 16,30, 33,23, 30,61, 62,45, 59,119, 116,90, 156,198, 373,326 classes=80 num=9 jitter=.3 ignore_thresh = .7 truth_thresh = 1 random=1

Thanks.

[ardianumam]

You can, e.g., take a look on the code that calls the frozen model (*.pb file). There should be corresponding tensor output names needed to call. Or, you can ask directly to the repo owner of the YOLO model you use.

[Diya2507]

import tensorflow as tf

tf.train.import_meta_graph("./model.ckpt-200000.meta") for n in tf.get_default_graph().as_graph_def().node: print(n) Once imported, you can make a summary file for Tensorboard, which allows you to visualize the graph nicely:

with tf.Session() as sess: writer = tf.summary.FileWriter("./output/", sess.graph) writer.close() To see the saved summary file in Tensorboard, run tensorboard --logdir=./output/

after getting the output name.

[sumanth13131]

I used your_outputs = ["dense_3/Softmax"] working fine i get it from model summary

[MuhammadAsadJaved]

@Diya2507 From the tensorboard these are details about all the model/graph. Which one is the output_names?

[MuhammadAsadJaved]

@DanielJean007 If you are using YOLOV3 then output nodes are:

output_node_names = ["input/input_data", "pred_sbbox/concat_2", "pred_mbbox/concat_2", "pred_lbbox/concat_2"]

See the reference

[MuhammadAsadJaved]

In object detection, we need at least 3 tensors to be called: (i) input, (ii) class prediction and (iii) box prediction. In this case, we need to know those tensor names when calling them. How to know those names? It depends who makes (who gives the tensor name). This repo uses model from here. I can know the tensor name outputs by seeing how this repo call frozen model to do prediction.

@ardianumam As you said you are using this repo for .pb and you can know about output_names from his prediction script. But why your and his output_names are different in the demo?

Your : your_outputs = ["Placeholder:0", "concat_9:0", "mul_9:0"] His : output_node_names = ["input/input_data", "pred_sbbox/concat_2", "pred_mbbox/concat_2", "pred_lbbox/concat_2"] or maybe: return_elements = ["input/input_data:0", "pred_sbbox/concat_2:0", "pred_mbbox/concat_2:0", "pred_lbbox/concat_2:0"]

Which one is right?