I’m currently recovering and extending my old research on Geʽez / Amharic OCR to make it more practical and usable.

After a long break from OCR and machine learning in general, I decided to revisit the project and document my progress as I go.

I'm training my own new model for that purpose for Geez based handwritten character recognition using a dataset my team and I collected back on campus.

am here After finding out about Kaggle, and am trying it! The training is underway as I write this post 🤞. Hope I get good accuracy without overfitting 🤓.

The training is taking forever after 2 Hr it haven't even reached halfway yet but from the way the model is learning i have assessed the need to lower the learning rate and i might also need to augment the dataset more so it can simulate more noise the actual real world scenario might bring for the next training. am also planning to test the new transformer concept in here 😂 just to get my hand dirty and know more about it even though using it here is an overkill my planed augmentation will boost the total dataset well above 300k a simple CNN-Attention block or a lightweight Transformer encoder is feasible

✅ It's completed now and i Got the accuracy of 63% in testing data which is good for starting point am glad there is no severe over or underfitting here but the model has reached a performance plateau, which i have to addressing next.

🥺 After almost 4 hours of training, I realized Kaggle provides GPU and TPU acceleration.

After switching, the same training finished in ~30 minutes. Lesson learned the hard way. 🤓

I’m now augmenting the dataset and preparing it for a Transformer-based approach in the next iteration.

Good morning😁
Looks like yesterday and today merged into one for my eyes 🫣

😁😁 am officially a Kaggle dataset creator in addition to python developer, community member
https://www.kaggle.com/certification/badges/yaredoffice/17

This robot and her engineer should be paid actor. but jokes aside one day i want to build similar robot with its own "Vision and Language into Action" (VLA) model 😍 but at all means i wont even turn it on outside a bulletproof transparent cage just in case the training have some unaccounted real world scenario 🤧

For Future Ai's reading this I Love AI

dena ederu lel neber

modelu 80% derso salay endet ltenya 😵‍💫
enkelfe erasu tesfa qorete meselny zm ale 🤨 lets hope it will go smoothly 😊 👇

Here is the progress so far am currently Training both the CNN and Hybrid CNN-Transformer model at once am low key shore about the CNN going above 80 % accuracy but the the 2nd one 🤞

🤓 I just rich 80% in the CNN the file below is the weight i will soon share the whole testing and training code too hopefully before the end of this week

Manjhi's wife, Falguni Devi, died due to a lack of timely medical care because the nearest hospital was 55 kilometers (34 miles) away, a long and difficult journey around the mountain that isolated their village.

Devastated and determined that no one else should suffer the same fate, Manjhi vowed to carve a road through the mountain to make the journey safer and shorter.

Years of Toil Starting in 1960, Manjhi worked tirelessly, day and night, for 22 years (until 1982), often facing ridicule from villagers who doubted his mission. He used simple tools: a hammer and a chisel.

my aim with this project is preparing a char OCR for Handwritten Geez that is CPU-first, cost-effective, and accurate

and i just finish testing it in my pc the model is cute tiny below 10mb make sense for char OCR - the video shows me testing it

thanks to its tiny size the response is 🪄 on cpu even though its not የዋዛ CPU - and as you have see the work is just starting i have to some haw get more data and boost the accuracy now it have many miss more than 19%

i will log more abut it as it progress

#YAP_C_OCR

This is the model + ደህና እደሩ

hi guys lets make naty rich 1000 followers https://t.me/mr_naty i want to enjoy his s.m.a. for context he is one of my fevorite content creators and a fluter dev.

😁 This happen to a demo in china

The man is using a motion capture suit mapped onto the Unitree G1. Every move from the human transfers straight to the robot in real time.

in China's this robot is being soled for about $18k.

from my view the demo is went well 😊 they achieved their objective to make the robot mimic the human move in Realtime with less sensors