🌱

Muh. Naufal

🇮🇩 Informatics Student • Mobile ML Developer

Hi, I'm Muh. Naufal!

Informatics student building Android apps that can see — using machine learning & computer vision.

🎯 3,500+ Dataset ⚡ <120ms Inference 📱 Offline-First
★

Lets Gooo! 🍃

Foto Muh. Naufal mengenakan topeng Spider-Man
Muh. Naufal 👨‍💻 UNAMIN Informatics 2025
🌱

About Me

3rd Year Informatics Engineering Student at Universitas Muhammadiyah Sorong (UNAMIN) 🎓

I'm a third-year Informatics Engineering student at Universitas Muhammadiyah Sorong (UNAMIN), focused on Android development and machine learning for image classification. I build apps that run the model directly on the phone, so they keep working in places with no internet, which is the everyday reality for many farms and villages in Papua Barat Daya.

My main project, Banalea, detects banana leaf diseases such as Sigatoka, Cordana, and Pestalotiopsis from a single photo, so farmers can catch problems early without waiting for an expert.

📍 Sorong, Papua Barat Daya 🚀 Open Source Contributor
🧰

Skills & Tech Arsenal

Every tool tested in the field and compiled on real devices

Mobile (Android) 📱
Kotlin CameraX Android Studio
AI / ML & Vision 🧠
TensorFlow Lite MobileNetV4 ResNet50 OpenCV PyTorch Image Classification Transfer Learning
Tools & Pipeline ⚡
Git & GitHub Figma Google Cloud Weights & Biases Google Colab Gradle Kotlin DSL
⭐ FLAGSHIP INNOVATION 2024

Banalea — Smart Banana Leaf Disease Detector

Instant on-device diagnostic camera classifying Sigatoka, Cordana, Pestalotiopsis, and Healthy banana foliage in milliseconds.

Offline-First ⚡ GitHub
My Role:

Lead Mobile Developer & ML Engineer — collected 3,500+ agricultural leaf dataset, trained dual MobileNetV4 & ResNet50 pipelines, optimized with TFLite for latency under 120ms on low-cost devices.

Latency: ~118ms Model Size: 4.2 MB

How It Works In The Field 3 Easy Steps

01
Take a Photo

Snap the affected leaf with CameraX auto-guide frame and real-time lighting detection.

02
On-Device Inference

Dual MobileNetV4 & ResNet50 run instant inference on-device without needing internet connection.

03
Instant Results

Get high-confidence diagnosis, severity score, and practical local organic remedy suggestions.

Classified Botanical Pathology

🍂
Black Sigatoka

Dark streaks along leaf veins caused by Pseudocercospora fijiensis.

⭕
Cordana Leaf Spot

Large oval spots with pale grayish center and bright reddish-brown border margins.

🥀
Pestalotiopsis

Necrotic spreading lesions starting from leaf tips, commonly caused by humidity distress.

🌿
Healthy Foliage (Sehat)

Vibrant emerald green surface with intact parallel veins and optimal photosynthetic rate.

Live Classifier UI Preview Active Inference
Banalea Hasil Pemindaian Mobile Screen
Real App Output: Hasil Pemindaian
Daun Sehat (Healthy Leaf)

On-device dual-backbone inference verifies healthy banana foliage with high confidence directly from field camera scans without network overhead.

MobileNetV4 Confidence 0%
ResNet50 Cross-Validation 0%
📋
Rekomendasi Petani: Tanaman dalam kondisi baik. Pertahankan kondisi lingkungan dan rutinitas perawatan yang sudah dilakukan.
🍃
📫 INBOX IS ALWAYS OPEN

Got an exciting project or want to collaborate?

Whether you want to chat about Computer Vision, Android development, or swap farming tech ideas, my inbox is always open!