Deep Learning & CNN using caffe/tensorflow & Machine Learning
$30-250 USD
Chiuso
Pubblicato circa 6 anni fa
$30-250 USD
Pagato al completamento
looking for an expert in deep learning neural network, machine learning and python, who is knowledgeable in alexnet, ResNet architecture and caffe, training the network and extracting key features and who also has solid background in SVM, image processing, face detection and recognition.
- must be capable of training and fine tuning cnn (Alexnet) neural network, extracting features for further training
must be capable of training svm classifier with features extracted from the pre-trained caffemodel or a given dataset ?
- must be capable of not only replace and retrain the classifier on top of the ConvNet on the new dataset, but to also fine-tune the weights of the pretrained network by continuing the backpropagation. It is possible to fine-tune all the layers of the ConvNet, or it’s possible to keep some of the earlier layers fixed (due to overfitting concerns) and only fine-tune some higher-level portion of the network.
- must be capable of building and compiling of caffe, python in linux (ubuntu)
- train and test models and improve the accuracy of the classification results in caffe (Alexnet), ResNet and etc.....
Skills required: Matlab, Python Deep Learning, Deep Neural Networks, Computer Vision, Image Processing, Computer Science
Must be able to execute & implement the document uploaded generating similar results but with a different dataset
I am here freelancer first to discuss the details of Deep Learning & CNN using caffe then
i can sure about my price and the deadline. My way of working is not only to complete but also to
provide enough understanding to the project owner. So this will be base of our long term relationships.
I am a Researcher with experience working with Alexnet as feature extractor for using it with other classification layers as could be SVM.
You can check my Google Scholar to see Machine learning related works (even tough I have works with AlexNet that have not been published yet) .