Dataset
The dataset consist of two label images:
- Meme
- Not Meme
Meme folder consist of 222 meme images and Not Meme folder consist of 108 non meme files. Meme file consist most of the images contaning the text on the picture and not meme consist of all type of images from sports to the text in various forms like document, image text to get the higher accuracy and understand about the meme in a most efficient way.
UseCase
- Content Moderation - The meme classification model can be used to filter out the content of meme from the vast amount of data generated for the specific domain from the social media for the better understanding.
Future Scope
- Further work on the sentiment of the meme image like positive, voilence, offensive, sarcasm, neutral, etc. This can be used for various task like:
- Education - To eliminate the offensive content from the curated memes for education
- Brand Monitoring - To understand the sentiments of the user by understanding the representation by meme culture for decision making process.
Model Trained Using AutoTrain
- Problem type: Binary Classification
- Model ID: 42897109437
- CO2 Emissions (in grams): 1.1329
Validation Metrics
- Loss: 0.025
- Accuracy: 1.000
- Precision: 1.000
- Recall: 1.000
- AUC: 1.000
- F1: 1.000