CLIP (Contrastive Language–Image Pre-training) training code for Bangla.

Live Demo: HuggingFace Space

---

Installation

• python >= 3.9
• pip install -r requirements.txt

Bangla CLIP

The model consists of an EfficientNet / ResNet image encoder and a BERT text encoder and was trained on multiple datasets from Bangla image-text domain. To start training,

python train_clip_bangla.py

---

Image Search Demo with Bangla CLIP

• Search App Code: bangla-image-search
• Article: medium

Citation

@misc{nazi2024bangla,
  author = {Zabir Al Nazi},
  title = {A dead-simple image search engine for Bangla using CLIP (Contrastive Language–Image Pre-training)},
  howpublished = {\url{https://github.com/zabir-nabil/bangla-CLIP}},
  year = {2023},
}