CineScan

🎬 Inspiration

We were inspired by the growing number of movies available online and how difficult it is for users to find relevant content. We wanted to use AI to detect and classify movies to make discovery faster and smarter.

🤖 What it does

CineScan automatically detects movie genres using both the poster image and description. It helps users identify what kind of movie they’re looking at, making browsing and organizing easier.

🛠️ How we built it

We collected data (movie posters, titles, descriptions) and used NLP to process text and CNN models to analyze images. We combined both results using a hybrid AI model and built a simple web interface with Flask.

🚧 Challenges we ran into

• Handling imbalanced genre data
• Training image and text models together
• Building a clean, responsive user interface
• Keeping prediction time low

🏆 Accomplishments that we're proud of

• Successfully merging two types of data (text and image)
• Building a working end-to-end AI system
• Learning new tools and deploying the model in a web app

📘 What we learned

We learned how to preprocess and model multimodal data, how to fine-tune machine learning models, and how to build and connect backend AI with a frontend interface.

🚀 What's next for CineScan

We plan to add user-based movie recommendations, support for multilingual content, and improve accuracy by training on larger and more diverse datasets.

Built With

• css-frameworks:-flask-(for-web-app)
• google-colab
• html
• languages:-python
• matplotlib-databases:-sqlite-(for-storing-movie-info)-apis:-tmdb-api-(for-movie-data-and-posters)-platforms/tools:-github
• nltk
• numpy
• opencv
• pandas
• tensorflow/keras-(for-ai-models)-libraries:-scikit-learn
• vs-code-cloud-services-(optional):-render/heroku-(for-deployment)