Questions tagged [tensorflow]

**IMPORTANT**: PLEASE ADD THE LANGUAGE TAG YOU ARE DEVELOPING IN. TENSORFLOW SUPPORTS MORE THAN ONE LANGUAGE. TensorFlow is an open source library for machine learning and machine intelligence. It is developed by Google and became open source in November 2015.

0
votes
0answers
3 views

Loading tensorflow object detection model post-training quantization to fp16 is extremely slow

Here I get a tensorflow object detection model through its provided object detection api. For some Nvidia GPUs (V-series, P100, ...) have supported float16 for faster training and inference, ...
0
votes
0answers
6 views

How tensorflow call CUDA apis to use gpu resources?

I'm new to tensorflow and now doing some experiments to place some tensorflow operations on pointed gpu device. But I'm just confused how it can call CUDA apis to achieve parallel computing. Does ...
0
votes
0answers
7 views

Tensorflow : How to use ResidualWrapper and HighwayWrapper in tensorflow?

I am trying to use ResidualWrapper and HighwayWrapper in my network structure but I am getting shape mismatch error. So What I have tried : import tensorflow as tf from tensorflow.contrib import rnn ...
1
vote
1answer
14 views

How to mask zero-padding values in Tensorflow Encoder-Decoder RNN with Attention?

In the official Tensorflow Neural Machine Translation example (https://www.tensorflow.org/alpha/tutorials/text/nmt_with_attention), in the Encoder model, a GRU layer is defined. However, the zero-...
1
vote
0answers
12 views

Looking for the proof, that Tensorflow noise generation can be unreliable on GPU

While reading the paper Rainbow from DeepMind, I got stuck on the following text in the "Experimental Methods" section: Tensorflow noise generation can be unreliable on GPU. That sounds surprising....
1
vote
1answer
10 views

Why does tf.keras.layers.Conv2DTranspose need no output_shape compared to tf.nn.conv2d_transpose?

I am missing something basic here. But I always used the tf.nn API for transpose convolution, where I have to specify the output shape, because it is ambiguous(Link). However, for TF 2.0 I switch to ...
-1
votes
0answers
10 views

How to draw lines with the output coordinates of bounding boxes?

I am using Tensorflow object detection api for detecting points in an madical image, and I have done the detection, now I want to draw 5 lines between some points, how should I do for getting the ...
0
votes
0answers
16 views

Issues splitting autoencoder model in Keras

I have trained an autoencoder and saved it using keras built in save() method. Now I want to split it into two parts: Encoder and decoder. I can successfully load the model and get the encoder part by ...
0
votes
0answers
16 views

Tensorflow: All components must have the same size in the 0th dimension

I have the following code: import tensorflow as tf filenames = # list of N filenames of input data labels = # NxM array where each row represents a data in one-hot-encoding form noise = # list of N ...
-2
votes
1answer
17 views

ubuntu 16.04 pyinstaller

ubuntu 16.04 pyinstaller 3.5 dev python 3.5 I am trying to create an executable file in Ubuntu using pyinstaller. If you run py file using pyinstaller -F, it will be created successfully. Then, when ...
-3
votes
0answers
7 views

Retrain InceptionV3 model with imagenet weights for additional data and classes [on hold]

I want to retrain the inceptionV3 model for some additional data, suppose if we pass the Shimla apple image to inceptionV3 for prediction it shows the result as Granny Smith which is a green apple. so ...
0
votes
0answers
14 views

an error with dimensions in CNN using tensorflow

I am doing a basic tensorflow driven CNN. There is some kind of dimension which I am unable to locate. thanks in advance I am working with jupyter in my system. I run on miniconda environment. pred =...
-2
votes
0answers
16 views

Object detection for already trained model

Currently done: I have trained my custom model. It classifies two types - first is when object X is present in the image, second not present. For object classification this is ok. Problem to solve: I ...
0
votes
0answers
8 views

Converting Darknet53 gives “nan” results in Tensorflow 2.0

I am trying to convert Yolo v3 to tensorflow 2.0 I wrote the Darknet53 network layers and I am able to run it on both test inputs as well as actual images but the result in both cases is 'nan' I ...
0
votes
0answers
15 views

when tensorflow-demo(detectorActivity) is implemented in android things, it says “session 0: failed to create capture session; configuration failed”

ive been trying to implement the tensorflow-demo detectoractivity in android things but black screen just appears. here is the logcat 2019-04-24 13:55:00.735 5046-5046/org.tensorflow.demo I/...
-4
votes
0answers
17 views

I want to convert handwritten dataset from PNG to CSV in Python [on hold]

I have a dataset for Handwritten digits as PNG format in two directories (train & test ) and i want to convert it to csv format For easy handling , this is a snapshot to my datasethandwritten ...
-4
votes
0answers
29 views

please explain the working of this python tensorflow code

it's a face detection attendance system project,which at a time recognizes one face and marks their attendance.But, can't understand the working of this code This requires retrained_labels.txt and ...
-2
votes
0answers
26 views

Tensorflow android app does not install on device

I followed exactly same instructions as mentioned in this tutorial. https://github.com/tensorflow/examples/blob/master/lite/examples/image_classification/android/README.md The (debug as well as ...
1
vote
0answers
20 views

Unable to use online prediction service with image input using a model saved from keras

Consider the following code to create a saved_model import tensorflow as tf from tensorflow.keras.layers import Input, Lambda from tensorflow.keras.models import Model inp = Input((), dtype=tf....
0
votes
0answers
17 views

Nonetype object has no attribute get tensorflow object detection from image

I am facing Nonetype Object has no attribute error I have the script which was working fine 3 days back but today it throws an error. Below is the code def recognize_object(model_name,ckpt_path,...
0
votes
0answers
6 views

How to resolve interactive session error in tensorflow (version 2.0.0a), python version 3.6.8, anaconda version 5.3.1?

I run the following code snippet from __future__ import print_function import tensorflow as tf import numpy as np which returns the following error `` AttributeError Traceback (most recent call ...
-1
votes
0answers
16 views

how do I solve this error AssertionError when I try to implement JPEG attack layer?

I have an autoencoder and I want to make jpeg attack as a noise layer. because jpeg quantization is not differentiable I used an estimation instead of it. my network has two input with shape ( channel ...
0
votes
1answer
33 views

Use neural network to learn distribution of values for classification

Use neural network to learn distribution of values for classification The aim is to classify 1-D inputs using a neural network. There are two classes that should be classified, A and B. Each input, ...
0
votes
0answers
10 views

how to use tf.contrib.framework.fuse_op to fuse graph

I want to use tf.contrib.framework.fuse_op to fuse graph to fuse my graph for faster inference,but when i call the api input_list = ["ac_input"] output_list = ["Inference/Reshape"] default_type = ...
0
votes
0answers
8 views

Making predictions with RetinaNet TPU

We may be overlooking something entirely, but my team is currently trying to train and make predictions using the RetinaNet TPU implementation on GCP. However, we are stuck trying to simply make ...
0
votes
0answers
24 views

Keras output shape is incorrect

I'm setting up a recurrent neural network to classify words between English or French and I'm getting an error about the output shape being incorrect. My network looks like this: ...
1
vote
1answer
21 views

The confidence level of tensorflow object detection API

I am working on performing object detection via tensorflow, and I am facing problems that the object etection is not very accurate. The architecture I am using is faster_rcnn_resnet_101. More ...
0
votes
0answers
15 views

Tensorflow Object Detection API: Inference pre-trained Kitti Model on Cityscapes Data leads to low accuracy

I am currently using the TF OD API. However, I run into some issues with the inference. Steps to Reproduce my issue: Download Faster R-CNN Resnet101 model pre-trained on Kitti Data Create Tf-Record ...
0
votes
1answer
21 views

Can a neural network be configured to output a matrix in Keras?

I am working with predicting a time series in 3 dimensions and would like to know if it is possible to configure a model to output a martix in Keras. Currently, I have 3 regression models I train ...
0
votes
0answers
7 views

How to add computation (eg log and sqrt root) to keras vgg16 before softmax layer

I am implementing the improved bilinear pooling (http://vis-www.cs.umass.edu/bcnn/docs/improved_bcnn.pdf) and I want to add computation (eg log and sqrt root) before softmax layer to my model, which ...
1
vote
1answer
13 views

Missing val_acc after fitting sequential model

I am missing information about the 'val_acc' attribute when I fit a compiled sequential model. I have a sequential model that is compiled with 'accuracy' metrics model = Sequential() model.add(...
0
votes
1answer
18 views

Why is the model not learning with pretrained vgg16 in keras?

I am using the pre-trained VGG 16 model available with Keras and applying it on the SVHN dataset which is a dataset of 10 classes of number 0 - 10. The network is not learning and has been stuck at 0....
1
vote
0answers
15 views

Using regularization with quantized graph in TensorFlow

I am trying to use Quantization-aware training for mobilenet_v1 on my own dataset. I first train a network with tf.losses.softmax_cross_entropy for the classification output, with tf.nn.l2_loss ...
0
votes
0answers
11 views

Input to DecodeRaw is not a multiple of 8, the size of double

I want to build a tensorflow dataset from tfrecords. this is my code: def make_dataset(): filenames = [train_tfrecords_dir + name for name in os.listdir(train_tfrecords_dir)] dataset = tf....
0
votes
0answers
12 views

TF 2.0 GPU cofingure, how to configure the tensorflow-gpu 2.0 and cuda, cudnn after create another virtual environment in pycharm

I have installed TensorFlow 1.12 in Pycharm using a virtual environment, the GPU can work well. When I create another virtual environment and install TensorFlow 2.0 Alpha, the GPU doesn't work anymore....
0
votes
0answers
3 views

Mute logging in TPU estimator

I am using BERT run_classifier and tensorflow TPUEstimator, each time I train my model or I predict using the estimator predictor, I get too many logging information printed on my screen. How can I ...
1
vote
1answer
21 views

How to put a list of values in a tensor at given positions with TensorFlow

I'm looking for a (the best ?) way to put a set of values in a tensor at given positions ; or an operation involving the tensor, the list of values and the list of indexes, and outputing the same ...
0
votes
0answers
10 views

Is this implementation of VQ-VAE for audio incorrect?

I was hoping to find an open-source implementation of the Neural Discrete Representation Learning paper for audio. I am looking at this github repository. The paper says that "the encoder has 6 ...
-1
votes
0answers
15 views

I need a “print digits” dataset to train model which will recognize digits from images. or advises. thanks

I tried to use tessaract to detect digits, but I had no success. samples of images which I need to be recognized
0
votes
0answers
14 views

How to see checkpoint tensorflow models in tensorboard

I want to use some pretrained model from the below page : https://github.com/tensorflow/models/tree/master/research/slim#pre-trained-models As I understand they are the code and the checkpoint file. ...
0
votes
0answers
10 views

Why can this RBF network converge to a linear function but not to a quadratic?

I'm writing an RBF neural network and have it predict the solution to a simple ODE. When the solution to the ODE is linear (say, $$ \frac{dy}{dx}=1)$$ then the model accurately predicts $$y=x+c$$. ...
0
votes
0answers
7 views

ImportError : libcublas.so.9.0 : no such file or directory

I have installed tensorflow (ubuntu with pip and also tried with anaconda) without GPU and when I try to run a code with tensorflow I've got the error message mentioned in the title. I do not have a ...
1
vote
1answer
24 views

How exactly is RMSE computed in tensorflow?

The rmse computed by tensorflow does not match with the rmse computed manually by me. The relevant code has been pasted below : # Train a linear regression model. tf.logging.set_verbosity(tf.logging....
0
votes
0answers
9 views

How to implement Pixel-Adaptative Convolutional Layer

The drawback with classical CNNs is that we need to multiple the number of filters, so that each filter will be activated by a different object (for example an umbrella will activate the 67th filter ...
-1
votes
0answers
7 views

How to do eager execution of the tensorflow object detection API?

I want to modify the Faster RCNN code of tensorflow object detection api. So I want to run eager execution to see the changes. I am using train.py in the legacy folder.
-1
votes
0answers
14 views

Exporting a keras tf 2.0 model to c++

I have trained a keras tensorflow model in python, I want to export the model in some manner and make it run in c++. I have found multiple ways and doing it and I don't know which is the current best ...
0
votes
0answers
17 views

Tensorflow from source works slower than version installed from pip

When I build Tensorflow 1.13 from source, I noticed than it work slower than tensorflow-gpu installed from pip GPU: nvidia gtx 1080 Driver: 410 Cuda: 10.0 Cudnn 7.5 I tried cudnn 7.3. I set path ...
0
votes
0answers
14 views

export a savedmodel pb from tensorflow's example retrain.py

I have come across this repository: https://github.com/ArunMichaelDsouza/tensorflow-image-detection that utilizes tensorflow's example code: https://github.com/tensorflow/hub/blob/master/examples/...
0
votes
0answers
28 views

nvidia-docker GPU in Docker Container

I'm trying to replicate work/experiments which require me to follow this particular tutorial on setting up Jupyter + Tensorflow + Nvidia GPU + Docker + Google Compute Engine. ' I'm able to ...
-1
votes
2answers
21 views

Tensorflow use trained model for detection

I'm working on a waste/garbage detector for a personal project. I rely on Tensorflow (in Python 3) to train my own dataset. I have a script that creates and trains a model from scratch. Then, I ...

http://mssss.yulina-kosm.ru