Files
radiacode/docker-compose.yml
Jacquin Antoine 0847a3fc80 Fix: CsI(Tl) non-linear response correction + detector calibration overhaul
Root cause of Am-241 misidentification: the Radiacode 103's CsI(Tl) crystal
shifts low-energy peaks upward (59.5 keV → 71.6 keV for Am-241) due to
non-proportional scintillation response. The model was trained on theoretical
peak positions and couldn't match the shifted real peaks.

Changes:
- Add inverse CsI(Tl) non-linear correction to inference pipeline
  (radiacode_monitor.py, web/config.py, test_detection.py)
  E_apparent = E_true * (1 + 0.37 * exp(-E_true/100))
  Corrects channel mapping so peaks appear at theoretical energies
- Fix energy calibration: DetectorConfig now uses E = 0.33 + 2.97*ch
  with 1023 channels, matching the real detector (was energy_min=20,
  skip_first_channel=True, different channel width)
- Add K-escape peaks for CsI(Tl) iodine X-ray escape (E - 28.5 keV)
- Add asymmetric peak shapes for low-energy tails (< 200 keV)
- Add log1p normalization in dataset and inference (replaces max-norm)
- Add background-subtracted training mode (subtract_background flag)
- Add low-signal augmentation (0.01-5 Bq activities, 30-300s durations)
- Update docker-compose.yml: batch_size=32, duration=30-300s,
  CSI_NONLINEAR_ALPHA/BETA env vars for detect and web
- Web dashboard: apply CsI correction to displayed spectra
- Various UI fixes (Chart.js width, zoom/pan, isotope lines)

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-21 17:35:22 +02:00

82 lines
2.1 KiB
YAML

services:
train:
build:
context: ./train
dockerfile: Dockerfile
volumes:
- ./data:/data
- ./models:/models
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: all
capabilities: [gpu]
environment:
- NVIDIA_VISIBLE_DEVICES=1
- CUDA_VISIBLE_DEVICES=1
- DATA_DIR=/data/synthetic
- MODEL_DIR=/models
- NUM_SAMPLES=50000
- EPOCHS=100
- BATCH_SIZE=32
- LEARNING_RATE=0.001
- DETECTOR=radiacode_103
- MIN_DURATION=30
- MAX_DURATION=300
- SEED=42
- MEASURED_BACKGROUND_PATH=/data/background_24h.npy
restart: "no"
detect:
build:
context: ./detect
dockerfile: Dockerfile
volumes:
- ./models:/models:ro
- ./logs:/logs
- ./data:/data
devices:
- /dev/bus/usb:/dev/bus/usb
privileged: true
environment:
- MODEL_PATH=/models/vega_best.pt
- ISOTOPE_INDEX_PATH=/models/vega_isotope_index.txt
- BACKGROUND_PATH=/data/background_24h.npy
- STATE_PATH=/data/monitor_state.json
- CPS_LOG_PATH=/data/cps_log.jsonl
- VEGA_ML_PATH=/models/vega_ml
- VEGA_DEVICE=cpu
- LOG_DIR=/logs
- SAMPLE_INTERVAL=60
- REPORT_HOUR=0
- MIN_LIVE_TIME=3600
- THRESHOLD=0.5
- CSI_NONLINEAR_ALPHA=0.37
- CSI_NONLINEAR_BETA=100.0
restart: unless-stopped
web:
build:
context: ./web
dockerfile: Dockerfile
ports:
- "8000:8080"
volumes:
- ./data:/data:ro
- ./logs:/logs:ro
- ./models/vega_isotope_index.txt:/models/vega_isotope_index.txt:ro
environment:
- STATE_PATH=/data/monitor_state.json
- CPS_LOG_PATH=/data/cps_log.jsonl
- BACKGROUND_PATH=/data/background_24h.npy
- BACKGROUND_SNAPSHOT_PATH=/data/background_snapshot.json
- LOG_DIR=/logs
- ISOTOPE_INDEX_PATH=/models/vega_isotope_index.txt
- ENERGY_CALIBRATION_OFFSET=0.33
- ENERGY_CALIBRATION_SLOPE=2.97
- CSI_NONLINEAR_ALPHA=0.37
- CSI_NONLINEAR_BETA=100.0
restart: unless-stopped