chore: remove sensitive thesis docs and notebooks from public repository

docs: add cloud multimodala fusion notebook for GPU inference

docs: finalize deployment troubleshooting guide and mark tasks 100% complete

fix: revert to native python:3.10-slim with CMAKE_BUILD_PARALLEL_LEVEL=1 to fix dlib numpy pybind ABI conflict

fix: enforce contiguous uint8 arrays for dlib and remove redundant rgb-bgr conversions

fix: correct GenConViT weight download directory path to be inside GenConViT package

fix: add missing albumentations dependency for GenConViT inference

fix: replace obsolete libgl1-mesa-glx with libgl1 for debian trixie

fix: replace python base image with miniconda to install precompiled dlib via conda and avoid OOM

fix: set MAX_JOBS=1 to prevent dlib C++ compilation OOM on HF Spaces

fix: quote dlib package string in Dockerfile to prevent shell redirection

fix: resolve OOM crash (CPU torch, isolated dlib) and Pillow 11 compatibility

fix: add all missing GenConViT deps (opencv-headless, dlib, face-recognition, decord, timm, torchvision)

fix: video import collision via importlib + auto-download GenConViT weights at startup

fix: pin gradio and starlette versions to bypass Starlette TemplateResponse breaking change

fix: monkey-patch gradio_client bool-schema crash at runtime before import

fix: replace gr.Code with gr.Textbox + show_api=False to fix bool-schema crash

fix: remove button_primary_shadow from theme (unsupported in Gradio 4.44+)

fix: pin huggingface_hub<0.26 to fix HfFolder crash, add .hfignore

Fix Hugging Face configuration and update docs

Merge branch 'main' of https://huggingface.co/spaces/Aryanaideveloper/Multimodal_Deepfake_Detection

Finalizing Thesis Multimodal Architecture for Hugging Face

initial commit

feat: implement fairseq compatibility shim and working audio inference pipeline

feat: initial project setup with audio/video detection modules and fusion