How rangilyu/nanodet works
NanoDet-Plus is a lightweight anchor-free object detector with a Python/PyTorch training and inference package and documented deployment demos. The main runtime path loads a configured model and checkpoint, preprocesses an image, runs detection, then visualizes results; training builds the model and computes losses through a task runner. The graph also preserves the documented COCO evaluation and ncnn, MNN, OpenVINO, LibTorch, and Android deployment branches. Some subsystem links rely on README or SOURCE INDEX references because their implementation excerpts were not provided.
Main entry points: inference.py Inference predictor · task.py Training task · nanodet_plus.py Detector architecture · nanodet_plus_head.py Detection head
Components 23
Data preparation
- Detection datasets
nanodet/data/dataset - Image transforms
nanodet/data/transform/pipeline.py - Batch processing
nanodet/data/batch_process.py
Detection model
- Detector architecture
nanodet/model/arch/nanodet_plus.py - Backbones
nanodet/model/backbone - Feature pyramid
nanodet/model/fpn/ghost_pan.py - Detection head
nanodet/model/head/nanodet_plus_head.py - Label assigners
nanodet/model/head/assigner/atss_assigner.py - Detection losses
nanodet/model/loss/gfocal_loss.py - Model building blocks
nanodet/model/module/conv.py
Training and inference
- Training task
nanodet/trainer/task.py - Optimizer builder
nanodet/optim/builder.py - COCO evaluator
nanodet/evaluator/coco_detection.py - Inference predictor
tools/inference.py - Model utilities
nanodet/util/check_point.py - Result visualization
nanodet/util/visualization.py
Deployment demos
- ncnn C++ demo
demo_ncnn/main.cpp - MNN C++ demo
demo_mnn/main.cpp - OpenVINO demo
demo_openvino/main.cpp - LibTorch demo
demo_libtorch/main.cpp - Android ncnn app
demo_android_ncnn/app/src/main/java/com/rangi/nanodet/MainActivity.java - Android native detector
demo_android_ncnn/app/src/main/cpp/NanoDet.cpp
Other components
- User
Connections 24
- User → Inference predictor: requests inference
tools/inference.py - Inference predictor → Detector architecture: builds model
tools/inference.py - Inference predictor → Model utilities: loads weights
tools/inference.py - Inference predictor → Image transforms: preprocesses image
tools/inference.py - Inference predictor → Result visualization: renders detections
tools/inference.py - Image transforms → Detection datasets: transforms metadata
- Image transforms → Batch processing: prepares inputs
- Detector architecture → Backbones: uses backbone
- Detector architecture → Feature pyramid: uses pyramid
- Detector architecture → Detection head: uses head
- Detection head → Label assigners: assigns labels
- Detection head → Detection losses: computes losses
- Detection head → Model building blocks: uses layers
- Training task → Detector architecture: builds model
nanodet/trainer/task.py - Training task → Batch processing: stacks images
nanodet/trainer/task.py - Training task → Optimizer builder: builds optimizer
- Training task → COCO evaluator: evaluates results
nanodet/trainer/task.py - User → Training task: trains detector
README.md - User → ncnn C++ demo: runs demo
README.md - User → MNN C++ demo: runs demo
README.md - User → OpenVINO demo: runs demo
README.md - User → LibTorch demo: runs demo
README.md - User → Android ncnn app: uses mobile app
README.md - Android ncnn app → Android native detector: invokes detector