https://github.com/vijishmadhavan/parse-clip
A simple CLIP based project for combining images from multiple datasets.
https://github.com/vijishmadhavan/parse-clip
clip data datacleaning dataexploration dataset fastai image python
Last synced: 3 months ago
JSON representation
A simple CLIP based project for combining images from multiple datasets.
- Host: GitHub
- URL: https://github.com/vijishmadhavan/parse-clip
- Owner: vijishmadhavan
- License: apache-2.0
- Created: 2022-03-18T13:52:06.000Z (about 3 years ago)
- Default Branch: master
- Last Pushed: 2022-03-19T09:57:28.000Z (about 3 years ago)
- Last Synced: 2024-12-30T16:53:02.384Z (5 months ago)
- Topics: clip, data, datacleaning, dataexploration, dataset, fastai, image, python
- Language: Jupyter Notebook
- Homepage:
- Size: 4.12 MB
- Stars: 3
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# PARSE-CLIP
A simple CLIP based project for combining images from multiple datasets. This has been very helpful for me, hope it helps others as well.
**Colab:** [
](https://colab.research.google.com/github/vijishmadhavan/PARSE-CLIP/blob/master/PARSE_CLIP.ipynb)

## What can PARSE-PIC do?
### - Combine Images from multiple datasets to create a new dataset.
#### Search Query - Carrot

### - Search the the class and get the total number.
#### Search Query - Indian
Total - 363 Images
-side.png)
### - Get the search results and explore.
#### Search Query - Red hair
-side.png)
### - Move Images to a new folder/drive and start training.
### - Remove unwanted Images from the dataset.
## Limitations
- It might struggle with huge datasets.
- Colab free version will be slow.
- We should have some idea about the dataset, random search wont work.
- Have tried it only on Kaggle datasets.## Acknowledgements
- [Beyond tags and entering the semantic search era on images with OpenAI CLIP](https://towardsdatascience.com/beyond-tags-and-entering-the-semantic-search-era-on-images-with-openai-clip-1f7d629a9978) by [Ramsri Goutham Golla](https://twitter.com/ramsri_goutham)
- [OpenAI's CLIP](https://github.com/openai/CLIP)
- [Natural Language Image Search](https://github.com/haltakov/natural-language-image-search)