Full Page Handwriting Recognition Model using Tensorflow
Budget: ₹1,500 – ₹12,500 INR
I am looking for a freelancer who can create and train a deep learning model that can accurately recognize handwritten text from paragraphs and documents.. The ideal candidate should have experience in the following:
- Deep Learning Model Creation and Training
- Image Recognition and Natural Language Processing
- Working with both image and text data
- Using TensorFlow as the preferred deep learning framework
- Transfer Learning
- The model will be trained on dataset with Hindi and Bengali Handwriting.
The project will involve the following tasks:
- Creating a deep learning model using TensorFlow for Full Page Handwriting Recognition
- The model will be based on the architecture used here: https://www.amazon.science/publications/sequence-to-sequence-contrastive-learning-for-text-recognition
- Resnet 152 (Initial weights can be taken from as the ImageNet.
- I will provide you with images and corresponding annotated text files (after the model creation is done).
- The images are of different sizes thus need to make a preprocessing file to bring all the images to same size before you input it to the model.
- Training the model using the prepared data
- Testing and evaluating the model's accuracy and performance
-The code should be modular and easy to understand. (Well Commented).
The candidate should have expertise in developing deep learning models and be able to deliver high-quality results within the given timeframe.
The deliverables include:
- Full Model code (working)
- The fine tunned resnet-152 weights and other weights of the fine tuned model.
I need the task to be done with priority ( Model creation should not take more than 3-4 days and training 1 day).
Budget: INR 3500-4000
- Deep Learning Model Creation and Training
- Image Recognition and Natural Language Processing
- Working with both image and text data
- Using TensorFlow as the preferred deep learning framework
- Transfer Learning
- The model will be trained on dataset with Hindi and Bengali Handwriting.
The project will involve the following tasks:
- Creating a deep learning model using TensorFlow for Full Page Handwriting Recognition
- The model will be based on the architecture used here: https://www.amazon.science/publications/sequence-to-sequence-contrastive-learning-for-text-recognition
- Resnet 152 (Initial weights can be taken from as the ImageNet.
- I will provide you with images and corresponding annotated text files (after the model creation is done).
- The images are of different sizes thus need to make a preprocessing file to bring all the images to same size before you input it to the model.
- Training the model using the prepared data
- Testing and evaluating the model's accuracy and performance
-The code should be modular and easy to understand. (Well Commented).
The candidate should have expertise in developing deep learning models and be able to deliver high-quality results within the given timeframe.
The deliverables include:
- Full Model code (working)
- The fine tunned resnet-152 weights and other weights of the fine tuned model.
I need the task to be done with priority ( Model creation should not take more than 3-4 days and training 1 day).
Budget: INR 3500-4000