My interests are Field Robotics, Applied machine learning, Large-language models, and AI-generated content (AIGC e.g., stable-diffusion).
- 3D object retrival tutorial: a toy example for 3D model retreival either from text or image, both using CLIP embedding and FAISS indexing
- GenAI game scene generation: Unity + 3D object retrival + diffuser + mono depth estimation demo (no code base available, but only sample playing)
- deepNIR: Dataset for generating synthetic NIR images and improved fruit detection system using deep learning techniques
- Fork of Imitation learning algorithms and co-training for Mobile ALOHA: many bug fixes, velocity feature added, more script for training, validation, and visualisation
- weedNet: Multispectral crop&weed dataset for semantic segmentation
- deepFruits: Kaggle hosted, 11 fruits bounding box annoration datasets
- gym rotors: Old (6yrs) but contains the fundamental toy examples for continuous RL control for a quadrotor
(updated on Nov/7/2025)
Kaggle is one of my favourite places to learn and explore new machine learning (ML) technologies. I have found it to be a fun and entertaining playground where I can apply my skills and knowledge to real-world problems.
If you are also interested in ML or other state-of-the-art (SOTA) technologies, I highly recommend checking out Kaggle. There are a wide range of challenging problems that need to be solved, and Kaggle provides a great platform for learning and collaboration.
(187/203,065 as of July/17/2025, Top 0.05%)
Here is my Kaggle profile page: My Kaggle profile
I hope you enjoy Kaggle as much as I do!