3.3 KiB
Diagnostic-app
The control station of the AIMsight diagnostic: see, analyze, launch. The Data page shows the curves, the camera images, the labels and the detectors' events; its Analysis tab launches, by claude -p, processes whose code lives in the other repositories:
- "Analyze the range": the
analystof the agents repository, on a chosen range, with the app's instructions (instructions/analyst.md) and without the detectors (DETECTORS: no): labels, submitted tolabeling/auto/, and a diagnostic report; - "Optimize the detectors", on a finished analysis:
/improvein Optimizer improves the detectors on the range of the analysis, its labels as truth, without overfitting; then thetool-builderof the agents repository brings the analyst's scripts into the reading harness (tools repository).
The architecture of the repositories, one responsibility each: docs/architecture.drawio.
The repository
app/: the web app (FastAPI server, React and Mantine page); its doc inapp/README.md.instructions/analyst.md: what the app asks of the analyst (output folder, report, dialogue).docs/: the architecture of the repositories, the plans.
And six links, ignored by git, to the neighbouring repositories (a single copy of each on the workstation):
| Link | Repository | What the app does with it |
|---|---|---|
data/ |
~/aimsight-data/data |
reads the data of the stations (outside git) |
labeling/ |
~/labeling |
reads and writes the labels; the analyses go in labeling/auto/analyses/ |
detectors/ |
~/detectors |
reads the detectors' events (scripts/categories.py) |
tools/ |
~/tools |
reads the data through tools/scripts/data.py; the analyst uses it, the tool-builder writes in it |
agents/ |
~/agents |
the plugin of the agents (--plugin-dir agents, --agent agents:analyst) |
optimizer/ |
~/Optimizer |
launches /improve there; reads the summaries in optimizer/sessions/ |
Each thing is committed in its own repository: the labels in ~/labeling, the detectors in ~/detectors (the optimization session does it, without pushing), the tools in ~/tools (the tool-builder does it).
Install and launch
ln -s ~/aimsight-data/data data && ln -s ~/labeling labeling && ln -s ~/detectors detectors
ln -s ~/tools tools && ln -s ~/agents agents && ln -s ~/Optimizer optimizer
The VS Code task Run app creates .venv and builds the front end if they are missing, then launches the server. By hand:
python3 -m venv .venv && .venv/bin/pip install -r app/requirements.txt
cd app/web && npm install && npm run build && cd ../..
.venv/bin/python app/server/server.py
The page opens on http://127.0.0.1:8770. claude (Claude Code, logged in to the subscription) must be on the PATH for the Analysis tab.
Pump the data
The pumping is in the ~/aimsight-data repository (extraction/, VS Code tasks pump and pump a range); user guide ~/aimsight-data/extraction/README.md.
Check
- the app:
.venv/bin/python app/server/check.py; - the labels:
python3 labeling/scripts/validate.py(the labels submitted by the app do not havereviewyet, which it requires: as many expected errors); - the detectors (benchmark, ~3 min):
python3 detectors/bench/regression.py --labels labeling/auto --data data.