Aminatron
Aminatron is a multi-object detection model.
It detects several objects on one image and returns object class, confidence, bounding box and an annotated image.
Folders
1 aminatron/
2 photos/ default input photos
3 model/aminatron.pt main Aminatron model
4 runs/predict/ output images with boxes
5 coco_yolo/ prepared COCO dataset
6 open_images_coco_ext_yolo/ prepared COCO + Open Images dataset
model/aminatron.pt is the main model everywhere. No yolo11s.pt or yolo11m.pt files are needed after aminatron.pt is created.
Install
1 cd /Users/aminmammadov/aiwork/models/aminatron
2 python3.12 -m venv .venv
3 source .venv/bin/activate
4 pip install -r requirements.txt
Create Aminatron Model
Create model/aminatron.pt from YOLO11m once:
python train.py --save-as-aminatronThis uses yolo11m.pt as source and saves it as:
Run On Laptop
Put images into:
Run:
Run on one custom image:
python predict.py path/to/image.jpgResults are saved to:
Prepare COCO
Prepare a smaller COCO subset:
python train.py --prepare-only --max-train 5000 --max-val 1000 --overwritePrepare full COCO:
python train.py --prepare-only --max-train 0 --max-val 0 --overwrite0 means no limit.
Fine-Tune Aminatron
Fine-tune the existing model/aminatron.pt:
python train.py --epochs 10Continue training later:
python train.py --epochs 5Both commands use model/aminatron.pt by default and save the best result back to model/aminatron.pt.
Add Open Images V7 Classes Without Forgetting COCO
Open Images V7 is useful for adding more objects like fruits, toys and animals. Do not train on Open Images alone if you want to keep old COCO classes. Prepare COCO first, then prepare a mixed COCO + Open Images dataset.
Prepare COCO if coco_yolo/data.yaml does not exist:
python train.py --prepare-only --max-train 5000 --max-val 1000 --overwriteInstall the Open Images preparation dependency locally:
pip install -r requirements-open-images.txtFiftyOne upgrades Pillow above the version required by Gradio. That is okay for dataset preparation/training. Before running the web app again, run pip install -r requirements.txt to restore the app dependencies.
Prepare a small starter subset:
python prepare_open_images.py --max-train 2000 --max-val 500 --overwritePrepare a tiny test subset first if you only want to check that the pipeline works:
python prepare_open_images.py --max-train 50 --max-val 20 --overwritePrepare only specific classes:
python prepare_open_images.py --classes Apple Banana Orange Toy Doll "Teddy bear" Ball Cat Dog Person --max-train 3000 --max-val 700 --overwriteThe result is saved to:
open_images_coco_ext_yolo/data.yamlThis mixed dataset uses:
1 coco_yolo/images/train + open_images_coco_ext_yolo/images/train
2 coco_yolo/images/val + open_images_coco_ext_yolo/images/val
Class ids are COCO-compatible. Matching Open Images classes are mapped into existing COCO ids, for example Person -> person, Dog -> dog, Apple -> apple, Teddy bear -> teddy bear. Truly new classes are added after the 80 COCO classes.
Because your previous test run produced an 18-class aminatron.pt, restore the COCO base before additive training:
python train.py --save-as-aminatronTrain Aminatron on COCO + Open Images together:
python train.py --data-dir open_images_coco_ext_yolo --no-auto-prepare --epochs 10--no-auto-prepare is intentional: it prevents train.py from accidentally preparing plain COCO if open_images_coco_ext_yolo/data.yaml is missing.
Important: old model files are backed up automatically to model/backups/ before model/aminatron.pt is overwritten.
Web Demo
Hugging Face Space Email Review
The Space can email the original uploaded image and the processed result image to:
mammadov.amin2000@gmail.comFor this to work, add these Hugging Face Space Secrets:
1 RESEND_API_KEY=your_resend_api_key
2 EMAIL_TO=mammadov.amin2000@gmail.com
3 EMAIL_FROM=Aminatron <onboarding@resend.dev>
Recommended setup:
Create a Resend account with mammadov.amin2000@gmail.com.
Create an API key at https://resend.com/api-keys .
Add RESEND_API_KEY to Hugging Face Space Secrets.
Keep EMAIL_TO=mammadov.amin2000@gmail.com.
If you use Resend without a verified custom domain, keep:
EMAIL_FROM=Aminatron <onboarding@resend.dev>This is better for Hugging Face Spaces than Gmail SMTP because SMTP ports can be unreachable from the Space container.
Fallback Gmail SMTP secrets, only if SMTP works in your environment:
1 SMTP_USER=your_gmail_address@gmail.com
2 SMTP_PASSWORD=your_gmail_app_password
3 EMAIL_TO=mammadov.amin2000@gmail.com
For Gmail, SMTP_PASSWORD must be a Gmail App Password, not your normal Gmail password.
You can also use this secret name instead of SMTP_PASSWORD:
GMAIL_APP_PASSWORD=your_gmail_app_passwordGmail setup:
Enable 2-Step Verification in your Google account.
Create an App Password for Mail.
Put that app password into SMTP_PASSWORD or GMAIL_APP_PASSWORD.
Optional secrets:
1 SMTP_HOST=smtp.gmail.com
2 SMTP_PORT=587
If SMTP secrets are missing, the Space still works, but photos are not emailed.
Most Needed Commands
1 cd /Users/aminmammadov/aiwork/models/aminatron
2 source .venv/bin/activate
3 python train.py --save-as-aminatron
4 python predict.py
5 python train.py --prepare-only --max-train 5000 --max-val 1000 --overwrite
6 pip install -r requirements-open-images.txt
7 python prepare_open_images.py --max-train 2000 --max-val 500 --overwrite
8 python train.py --save-as-aminatron
9 python train.py --data-dir open_images_coco_ext_yolo --no-auto-prepare --epochs 10
10 python train.py --epochs 10
11 python train.py --epochs 5
12 python app.py
