Repo containing some code and model examples from computer vision and NLP projects I have completed.
In the computer vision folder, we have the following projects:
- Microscope_slide_cancer_cell_classification - image classifer using edge detection to predict the type of cancer cell in a microscope slide
- Microscope_slide_cell_segmentation_tensorflow - image segmentation model to identify the location of cell nuclei within a microscope slide to aid in the prediction of cancer type
- Pipeline_optimisation_for_image_augmentation_tensorflow - optimising a tensorflow image pipeline to efficiently apply image transformations and augmentations to image datasets. Reduced time taken to process 100,000 images from 34 minutes down to 2 minutes.
- Road_location_satellite_GAN_tensorflow - Using a GAN to look at satellite images of a road and then create an image mask highlighting the pixel location of the road passing through the image. Used to identify road features and predict a road safety score.
- Land_use_satellite_classifier_tensorflow - image classifer to determine from satellite images if the area next to a road is in an urban or rural area. This was used to generate a safety score for a particular stretch of road.
- Roman_numeral_classification_tensorflow - image classifier to identify handwritten roman numerals Scenery_classifier_tensorflow - image classifier using transfer learning to determine which of 6 classes is being shown in a photograph of a natural or urban area.
- Synthetic_image_generation - using command line tools to generate synthetic images for a data driven machine learning competition
In the NLP (Natural Language Processing) folder, we have the following projects:
- Essay_grader_tensorflow - Building a tensorflow NLP model to parse an essay written by a student and predict the grade which the teacher would award for a certain essay.
- Movie_review_sentiment_classification_tensorflow - building a tensorflow model to perform sentiment analaysis on movie reviews and determine if the review is positive or neagtive in nature.
In the other section:
- Audiobook_marketing_prediction_tensorflow - predicting if a company customer is likely to be a return customer based on website usage patterns