Predict social media engagement before you publish. Engagement Prophesier is an AI-powered image engagement predictor for social media. It enables marketers to select the most engaging image from a set of options by analyzing visual and metadata attributes using a deep learning model.
Social media is saturated with content, and choosing the right image can significantly impact audience engagement. Engagement Prophesier uses machine learning to recommend the best-performing image from a set of candidates, eliminating human bias and guesswork.
- 📊 AI-driven prediction of likes and engagement levels for each image
- ⚖️ Bias-free selection based on data insights, not subjective judgment
- 💸 Revenue optimization by increasing post performance
- ⏱ Time-saving tool for marketing/designing teams
Upload candidate images, add hashtags and captions, and hit "Predict".
Each image receives a predicted engagement score and like count.
Image B gets ~30x more likes than Image A – visuals matter!
- Framework: FastAPI
- ML Libraries: Keras, NumPy, Pandas
- Model: CNN for image features + Fully Connected Layer for metadata
- Deployment: AWS Kubernetes cluster
- Framework: React.js
- 🎨 Frontend Repo
- Upload up to 4 images
- Add captions and hashtags
- Select publishing account and date
- Predict engagement per image
- Post the best one
- Priyanka Barnwal
- Prashant Gupta
- Pradeep Gowda
- Debojit Banik
Built lika a POC for a hackathon as part of the KHoders initiative at Khoros, Bangalore.