mrdbourke / nutrify

Take a photo of food and learn about it.

https://nutrify.app

mrdbourke/nutrify Issues

Create a way to generate images to boost inaccurate classes
Updated a year ago1
Datasets/papers to explore
Updated a year ago
Dashboard creation for data exploration
Updated a year ago7
Start breaking down `class_names` into more easily understandable foods
Updated a year ago2
Go through `all_images` on Google Storage and resize them to take up less space
Updated a year ago1
Change `autocorrect_and_merge_labels` pipeline to use OpenCLIP rather than generic CLIP
Updated a year ago
Autolabelling object detection and semantic segmentation
Updated a year ago
Autolabelling: Setup a data collection pipeline (e.g. what happens when new data comes in?)
Updated a year ago4
Create `export.py` (or something similar) for exporting models from PyTorch -> CoreML
Updated a year ago1
Update `README.md` for `foodvision` directory January 2023 overview
Closed a year ago2
Update main `README.md`
Updated a year ago
Creating a `config.py` or `config.yaml` for values of hyperparameters
Closed a year ago4
Add `image_source` tags to all images in the database
Updated a year ago
Simplify training scripts (e.g. merge `engine.py` -> `train.py`)
Updated a year ago
Clean scripts and add docstrings to code in `foodvision`
Updated a year ago
Clean/remove duplicate images with `fastdup`
Updated a year ago3
Move all computer vision code into `foodvision`
Closed a year ago2
Profile the training/testing code to see where the bottlenecks are
Updated a year ago
Only validate once per X epochs
Updated a year ago
Try using `polars` for fasting data loading or even use Hugging Face Datasets
Updated a year ago
Remove duplicate images from dataset (train and test)
Closed a year ago1
Use Cleanlab to clean labels in dataset
Updated 2 years ago
Create a database (PostgreSQL?) for labels
Updated 2 years ago1
Create a script to import the "100 most wrong predictions" and then export them as labelling task in Label Studio
Updated 2 years ago2
Add a log for each time a prediction is made (e.g. "pred_on_image": <INSERT_IMAGE>)
Updated 2 years ago1
Make a page for uploading data to Nutrify
Updated 2 years ago2
Create a function/script to create different models for different purposes (e.g. big, medium, small)
Updated 2 years ago
Add a way to save to storage the images uploaded to Nutrify
Updated 2 years ago3
Add a button for asking "is this correct?"
Closed 2 years ago4
Make the loading images function faster (this should appear on page load)
Updated 2 years ago2
Database update: make it so a new food gets added via FDC ID -> information pulled into database
Updated 2 years ago2
Model/class names not lined up + some classes are missing FDC data ("Egg Tart", "Fries", "Hamimelon")
Updated 2 years ago3
Build a food/not food model into the app
Closed 2 years ago2
Can't upload image
Closed 2 years ago3
Downloading images from Bing API only downloading first portion of string
Updated 3 years ago2
A domain.
Closed 3 years ago2
Mobile App
Closed 3 years ago2
Pytest and automated testing
Closed 3 years ago1
Change uploading size for an image
Closed 3 years ago1
Capturing Image on mobile & page didn't refresh after upload
Closed 3 years ago7
`pre-commit` breaking (not letting commits happen)
Closed 3 years ago4
Github actions
Closed 3 years ago3
Dependancy management with `pipenv` and `requirements.txt` and reducing build time
Updated 3 years ago
Documentation for setting up the project
Updated 3 years ago
Cleaning up and structuring the project
Updated 3 years ago
pytest, linting, code formatting, pipenv
Closed 3 years ago2
No "Successful" image uploaded
Closed 3 years ago1
Supressing error message for empty upload
Closed 3 years ago
Malicious content
Updated 3 years ago3