Comet.Box 📦📦

Comet.Box: A collection of efficient and fast implementation of SOTA Object Detection,Segementation and NLP Models.

Why Comet.Box?

• Want to get Started with Advanced Computer Vision and NLP?

  • Comet.Box provides you a large variety of Computer Vision and NLP pipelines.

• Want to run high-end, the latest state-of-the-art models, but you lack GPU and RAM?

  • All implementation in Comet Box is done using Google Collab or Kaggle Kernels

Some key Features!!

* Run training and inference in one place.
* All the implementation of NLP tasks is done by using Hugging Face.
* Train your model over customer data very easily.

Who can use the Comet.Box?

🚩 Comet.Box templates are meant for those people who have intermediate knowledge of Object Detection, Segementation models, and Algorithms.

🚩 Comet.Box only provides you a basic template to kick start your initial part on any Detection, Segementation, Gans, NLP project or task. You would still need to add some custom cfg and other files according to your datasets or corpus.

Important Elements

• A) Inference Engine
  • Original pre-trained models (from original authors and implementations) for inferencing and analyzing.
  • Pretrained models on coco, voc, cityscpaes, type datasets. It- Useful to analyze which algorithm works best for you.
  • Useful to generate semi-accurate annotations (coco, pascal-voc, yolo formats) on a new dataset.

Create real-world Object Detection applications

Wheat detection in field	Multi-Human Pose Detection	Ship Detection in Oceans on Satellite Images
Underwater Fish Detection	Real-Time Object Detection on Crowd	Real world Object Detection using Facebook DETR
Car Detection using YoloV5	Object Detection using RetinaNET	Human Pose Detection using HRNET
Mask Detection using CNN	Tiger Detection in wild using Yolov5

Gans and Transformer Applications

DeepFakes using GANS and AutoEncoders	CGans on MNIST

Real world Segmentation Models

Real Time Segmentation on Videos using YolACT++	Medical Images Segementation using UNET	Object Segmentation using Mask-RCNN

Classification Model's

COVID-19 Radiography Classification using Resnet18

Natural Language Processing

Sentence Classification using BERT Hugging Face	Question/Answering using DistillBERT

🧿Visit Model Zoo DET for Detection Models👓

🧿Visit Model Zoo SEG for Segmentation Models👓

🧿Visit NLP Model Zoo for NLP SOTA Models👓

🧿Visit Classification Model Zoo for Classification Models👓

Want to Contribute or join us?

🚀 To contribute to Comet.Box repository raise an issue in the git-repo or dm on LinkedIn

• Ritesh - Ritesh

  📧 To Join us Shoot a mail to

  • Mail Here

Upcoming Updates

• [ ] Detection Project:
  • [ ] ANPR
• [ ] Segmentation using:
  • [ ] Mask-RCCN
  • [ ] Detectron-2
• [ ] NLP Models
  • [x] Fine Tune BERT
  • [ ] Elmo Model
  • [x] DistilBERT
  • [ ] GPT-1 and GPT-2
• [ ] Object Tracking

We would love to hear from you about your feedback!! 🥰

• Drop us a Mail

Contributors ✨

Thanks to these wonderful people