deeplens.space

Mission Board Community

REAL SKY DATA. GAME-GRADE MOMENTUM.

Make real astronomical exploration feel immediate, visual, and worth coming back for.

DeepLens turns NASA survey data into a guided mission loop: choose a target, run a scientific scan, compare multiple processed views, and publish discoveries the community can verify.

Start a Mission See Community Finds

NASA Real survey data

4+ Scientific views per scan

XP Retention loop built in

Real NASA Data. Instantly Understandable.

Preview how DeepLens transforms raw telescope data into clearer scientific views before your first scan.

Survey: DSS2 Blue Pipeline: Deep Analysis Example: scan_1b2c97e45450

RAW ENHANCED

Preview media unavailable. Showing static DeepLens output from a real processed scan.

Mission Queue Loading guided targets…

Pick a highlighted region or roam freely across the sky.

Target Lock Awaiting target lock

Choose a mission or move the map to set coordinates.

Latest Scan No scan yet

Your next Deep Scan result will appear here.

Community Pulse Community feed is live

Discoveries, verifications, and ranks update below.

Onboarding + Mission Board

Guide new explorers into action fast.

Keep the flow obvious: learn the loop, pick a target, and get to the first Deep Scan with confidence.

How It Works

Navigate the sky or pick a mission

Hit Deep Scan

Explore views & analysis

Submit & share discoveries!

Missions — Pick a target to investigate

Observation Deck

Make the scan action feel like the core ritual.

The map, overlay, and action bar should feel like one focused command center rather than separate blocks.

DEEP SCAN IN PROGRESS

Fetching NASA survey data…

Usually takes 5-10 seconds

Overlay: 50%

Point the map at any sky region

Analysis Lab

Turn every scan into a strong result page.

Results should feel cinematic, trustworthy, and immediately actionable for both discovery and learning.

Submit a Discovery

One submission per scan — 3 free discoveries/month

Leaderboard

#	Explorer	Title	XP	Scans	Discoveries

Recent Discoveries by the Community

How DeepLens Works

Full transparency — we show you exactly where the data comes from, how we process it, and what every result means. No black boxes.

1 Where does the data come from?

When you click Deep Scan, our server sends a request to NASA SkyView — a real astronomical database that holds images from dozens of space telescopes and ground observatories. We download the raw data file (called a FITS file) for the exact sky coordinates you’re looking at.

This is the same data professional astronomers use. It’s not a photo — it’s raw telescope sensor readings.

2 How do we process it?

The raw data from a telescope looks mostly dark and hard to read. Our server runs several image processing algorithms to make features visible:

Enhanced view: Stretches the brightness so faint objects become visible (like turning up brightness, but smarter).
Heat map: Colors the image by intensity — bright areas glow orange/white, dim areas stay dark.
Structure detection: Highlights edges and boundaries — great for seeing galaxy arms and nebula shapes.
Denoised: Removes “static” noise from the detector, revealing hidden faint structures.

Raw

Enhanced

Heat Map

Structures

3 How does it identify objects?

After processing the image, we cross-match the coordinates with two professional astronomy databases:

SIMBAD (Strasbourg, France) — catalog of 15+ million known astronomical objects.
NED (NASA/Caltech) — catalog focused on galaxies and objects beyond our Milky Way.

If your scanned region contains a known object, we’ll tell you its name, type, and distance. If nothing is known — that’s exciting! It means you might be looking at something not yet cataloged.

4 What is the Novelty Score?

The Novelty Score (0-100%) tells you how “new” your scan region is. A high score means fewer known objects were found near your coordinates — combined with interesting features in the image, this suggests you may have found something worth investigating.

It’s not a guarantee of a discovery — it’s a scientific indicator that says “this area deserves a closer look.”

5 What is Quality Assurance (QA)?

Every scan goes through automatic quality checks:

SSIM check: Compares the processed image to the raw data to make sure processing didn’t create fake features.
Multi-spectral evidence: Checks X-ray (Chandra telescope) and radio (NVSS survey) catalogs to see if your region has signals in other wavelengths — real objects usually appear in multiple wavelengths.

6 Can I really discover something new?

Yes. The universe is enormous. Even with decades of surveys, vast regions remain unexplored in detail. Citizen scientists have discovered comets, supernovae, and even new types of galaxies. DeepLens gives you the tools that were previously available only to researchers with university telescope access.

Every discovery you submit is saved, gets a unique ID, and can be verified by other users. If verified, it becomes part of the DeepLens community catalog.

7 How do I submit a discovery & get my certificate?

Requirements: You must be signed in with Google and have run at least one Deep Scan. After the scan completes, the Submit a Discovery form appears below the results.

Auto-fill assistant: DeepLens generates a suggested classification, scientific write-up, and checklist based on the scan — click Fill everything to insert them automatically, or edit freely.
Free tier: 3 discovery submissions per month (resets on the 1st of each month). One submission per scan.
Discovery certificate: As soon as your submission is saved you receive a downloadable PNG certificate with your name, the classification, sky coordinates, date, and a unique certificate ID — ready to share on social media.
Discovery ID: Every submission gets a permanent unique ID you can use to track its verification status.

8 How does community verification work?

Any signed-in user can verify discoveries made by other explorers. Verification means you scan the same sky region and agree the object or anomaly is real.

How to verify: Find a discovery in the Recent Discoveries feed and click the Verify button. Each verification earns +5 XP.
Confirmed status: A discovery needs 30 independent verifications to reach Confirmed status and join the permanent community catalog. The submitter earns +100 bonus XP when this happens.
You cannot verify your own discoveries — independent confirmation is what makes the science credible.
Disagree option: If you scan the same region and the anomaly is not visible, you can cast a negative vote which reduces the verification count.

9 XP, Titles & the Astronomer Certificate

Every action on DeepLens earns experience points (XP) that unlock scientific titles:

Deep Scan → +10 XP
Submit a discovery → +25 XP
Verify another explorer’s discovery → +5 XP
Your discovery reaches Confirmed → +100 XP

Title progression:

0 XP — Explorer (starting title)
50 XP — Pioneer
150 XP — Scientist
300 XP — Astronomer — unlocks the Astronomer Certificate: a shareable certificate confirming your status as a community scientist
500 XP — Professor
800 XP — Director

Achievements (badges) are awarded automatically for milestones: first scan, 10 scans, 50 scans, first discovery, 5 verifications given, novelty score > 0.8, and first confirmed discovery.

Data Acquisition NASA SkyView

We query the NASA SkyView Virtual Observatory using the astroquery.skyview Python module. Default survey: DSS2 Blue (Digitized Sky Survey, 2nd generation). Image size: 600×600 pixels. Data format: FITS (Flexible Image Transport System) with WCS (World Coordinate System) headers.

Image Processing Pipeline Scientific

All processing uses standard astronomical techniques implemented with NumPy, SciPy, scikit-image, and Pillow:

Normalization: Percentile-based clipping (0.5%-99.8%) to handle astronomical dynamic range.
Enhanced: Asinh stretch (mimics human eye response to brightness) + unsharp masking.
Heat map: Normalized data mapped to a thermal color palette (black → red → yellow → white).
Structures: Sobel edge detection (scipy.ndimage.sobel) on both axes, combined as sqrt(Gx² + Gy²).
Denoise (NLM): Non-Local Means algorithm (skimage.restoration.denoise_nl_means) with h=1.2σ, fast_mode=True.
Background removal: 8×8 tile grid with sigma-clipped median + Gaussian interpolation background model.
Wavelet enhancement: À trous wavelet decomposition (B3-spline kernel) at 4 scales with per-scale gain boosting.
Frequency filtering: FFT-based bandpass filtering to suppress periodic detector artifacts.

Cross-Matching Engine Astroquery

Catalog queries use astroquery (Python package maintained by the Astropy Project):

SIMBAD: Query within 5 arcmin radius. Fields: object type (otype), V-band flux, distance modulus, angular separation.
NED: NASA/IPAC Extragalactic Database query. Returns galaxy types, redshifts, and distances.
Chandra X-ray: HEASARC catalog query for X-ray point sources.
NVSS Radio: VizieR catalog query for 1.4 GHz radio sources.

All catalog queries run concurrently (ThreadPoolExecutor) with a 10-second timeout per service.

Field Analysis Automated

The analyze_field() function computes:

Bright source count: Thresholding at mean + 5σ, connected component labeling.
Structure score: Normalized Sobel energy — how much edge content exists (0.0 – 1.0).
Dynamic range: Ratio of 99th to 1st percentile intensity.
Interest level: HIGH if ≥3 sources or structure ≥0.3; LOW if 0 sources and structure <0.1.
Tags: Automatically assigned based on metrics (e.g., “extended_emission”, “point_sources”, “high_contrast”).

Tech Stack Open Source

Backend: Python 3 + FastAPI (async web framework)
Astronomy: Astropy, Astroquery, Lenstronomy
Image Processing: NumPy, SciPy, scikit-image, Pillow
Sky Map: Aladin Lite v2 (CDS, Strasbourg) — pinned to local static snapshot for stability
Auth: Google Identity Services (JWT verification, server-side validation)
Frontend: Vanilla JavaScript (no framework, no build step)
Hosting: Self-hosted VPS with HTTPS (TLS via sslip.io certificate)
Storage: JSON flat files (users.json, discoveries.json) with thread-safe RLock writes
CORS: Configurable via DEEPLENS_CORS_ORIGINS environment variable

Available Processing Pipelines 8 Pipelines

Beyond the default Denoise pipeline, DeepLens offers seven additional specialized pipelines, each optimized for a different scientific scenario. The pipeline is selected at scan time via the POST /process API.

Denoise (Default): Non-Local Means algorithm (skimage.restoration.denoise_nl_means, h=1.2σ, fast_mode=True). Best general-purpose choice — removes detector noise while preserving faint extended structures.
Background Noise Removal: 8×8 tile sigma-clipped median background model with Gaussian interpolation. Ideal for regions with strong gradient backgrounds, such as near the Galactic plane or rich star fields.
FastAstro: Accelerated pipeline combining aggressive CLAHE normalization and Sobel edge detection. Trade-off: fastest results with slightly less sensitivity to extended emission. Good for rapid surveys.
JWST Denoising: Non-Local Means parameters re-tuned for JWST-like data characteristics — higher dynamic range, finer spatial resolution, and lower read noise model. Use for deep field targets.
JWST Pipeline: Full multi-step JWST-style processing: background subtraction, cosmic ray rejection simulation, percentage-clipped normalization, and asinh stretch. Best for long-exposure deep field analysis.
Noise2Noise Astro: Pair-based noise estimation using spatial bootstrap of the image itself. Particularly effective at revealing sub-threshold point sources in noisy fields where standard denoising blurs faint objects.
LIGO Noise Suppression: Frequency-domain bandpass filtering (FFT-based) adapted from gravitational wave signal processing techniques. Removes periodic detector artifacts most visible in calibration and structured fields.
Custom: User-configurable parameters — adjust percentile clipping bounds, stretch function, enhancement gain, and denoising strength via API body parameters.

Gamification & Ranking System Live Data

The leaderboard and XP system are fully server-side, tamper-resistant, and recalculated from verified scan and discovery events stored in the database.

XP events (server-side): Scan completed (+10 XP), discovery submitted (+25 XP), peer verification given (+5 XP), discovery confirmed by community (+100 XP to submitter).
Title promotion: Titles are recalculated automatically on every XP event — Explorer (0), Pioneer (50), Scientist (150), Astronomer (300), Professor (500), Director (800).
Leaderboard API: GET /leaderboard?limit=20 — sorted by XP, includes title, scan count, discovery count. Auto-refreshes every 60 seconds.
Personal profile API: GET /users/{username}/profile — full XP history, achievements unlocked, scan history, and monthly usage stats.
Monthly limits (free tier): Anonymous: 5 scans/month. Signed-in: unlimited scans, 3 discovery submissions/month, 15 Prof Aviv questions/month, 5 Prof Aviv chat sessions/month.
Achievements (7 badges): First Light (1st scan), Explorer (10 scans), Surveyor (50 scans), Sharp Eye (1st discovery), Peer Reviewer (5 verifications given), Catalog Buster (novelty ≥ 0.8), Confirmed Discovery (1 confirmed).
Confirmation threshold: 30 independent verifications promote a discovery to Confirmed status and trigger the +100 XP bonus.

Our Transparency Promise

DeepLens is built on one principle: you should never have to take our word for anything. Every piece of data is traceable, every algorithm is explainable, and every result can be independently verified.

Data Source Verification

✓ Every image comes directly from NASA SkyView — we don’t generate or modify raw data.

✓ Coordinates are shown for every scan — you can verify in any other sky viewer (Aladin, WorldWide Telescope, Stellarium).

✓ The “Raw Data” view shows the unmodified telescope image before any processing.

✓ Every scan report includes the exact survey name and pixel dimensions.

Processing Integrity

✓ SSIM Quality Check runs on every scan — it mathematically proves the processed images still match the original structure.

✓ All image processing uses published, peer-reviewed algorithms (Non-Local Means, Sobel, CLAHE, wavelets).

✓ No AI “hallucination” — we don’t use generative AI to create images. Every pixel comes from real telescope data.

✓ Processing parameters are fixed and deterministic — same input always produces same output.

Catalog Cross-Match Honesty

✓ SIMBAD and NED are maintained by professional astronomical institutions (CDS Strasbourg, NASA/Caltech).

✓ If an object is already known, we tell you — we don’t hide matches to inflate novelty scores.

✓ Novelty score formula is fully documented (see Technical Details tab).

✓ Multi-spectral QA checks X-ray and radio catalogs independently — real objects leave traces across wavelengths.

Discovery Verification

✓ Every discovery gets a unique ID and timestamp.

✓ Other users can independently verify discoveries by scanning the same coordinates.

✓ Verification requires multiple independent confirmations before a discovery is marked “verified.”

✓ We never delete or hide submitted discoveries — even unverified ones stay in the public record.

What DeepLens Does NOT Do

× We do NOT use AI to “imagine” or generate celestial objects. Every pixel is real telescope data.

× We do NOT claim processing artifacts are discoveries. Our SSIM quality check is specifically designed to catch this.

× We do NOT modify the coordinates or pixel values of raw data in any way.

× We do NOT sell user data, scan history, or discovery submissions to third parties — ever.

× We do NOT guarantee every high-novelty region is a genuine discovery — it is a scientifically grounded starting point for investigation, not a confirmed result.

× We do NOT delete submitted discoveries — even unverified or rejected ones remain in the permanent public record.

× We do NOT require payment to participate in citizen science — scanning, verifying, and the leaderboard are free for all signed-in users.

Your Data & Privacy

✓ Google Sign-In only stores your display name and email. No password is ever handled by DeepLens. Your Google token is verified server-side and never logged.

✓ Scan coordinates are public. Any scan you run — including anonymous scans — may appear in the community feed if a discovery is submitted from that scan.

✓ Discoveries are permanently public. When you submit a discovery, your display name and the sky coordinates are visible to everyone. This is intentional: open science requires open records.

✓ Monthly usage counters (scan count, discovery count, Prof Aviv questions) are stored server-side per user. They reset on the 1st of each month and are only used to enforce free-tier limits.

✓ Scan history is stored in your browser localStorage (last 20 scans) and also server-side in your user profile. You can clear browser storage at any time without affecting your leaderboard rank.

✓ No advertising, no tracking pixels, no analytics third parties. The only external services loaded are Google Sign-In (accounts.google.com) and jQuery from jsDelivr.

✓ To request deletion of your user profile or all associated discoveries, email us at hello@deeplens.space with your Google display name. We process deletion requests within 7 days.

Open Science & Community Input

DeepLens is built on the principle that science improves when more people participate. We welcome contributions, corrections, and criticism from the community.

✓ Pipeline suggestions: If you know of a better denoising algorithm, a useful catalog we don’t query, or a processing technique that could improve detection — tell us. We study every suggestion seriously.

✓ Bug reports & false positives: If a scan produces clearly wrong results, a broken image, or a misleading novelty score, please report it with the scan ID. We will investigate and fix it.

✓ Mission targets: Have a specific sky region you think deserves community attention? Suggest it as a mission target and we’ll consider adding it to the mission list.

✓ Scientific feedback: If you are a professional astronomer or researcher and you spot something worth following up, we actively want to hear from you. Credible leads are escalated to the admin team.

✓ Feature requests: New pipeline, new catalog, new UI feature — open a conversation via email and describe what you need.

Reach us at: hello@deeplens.space

Response time: typically within 48 hours. For urgent issues (server down, data corruption), mention “URGENT” in the subject line.

Questions? Doubt something? Found an error?

Good. Skepticism is the foundation of science. If anything looks off — a suspicious image, an unexplained result, or a score that doesn’t make sense — verify it yourself:

→ Aladin Lite — enter the RA/Dec coordinates to view the same region in any survey
→ SIMBAD — search for known objects at those coordinates
→ NED — NASA extragalactic database for galaxy and redshift data
→ NASA SkyView — download the same raw FITS data we use

That’s exactly why we show coordinates on everything. We want you to be able to audit every result independently.

Still have questions after checking? Contact us: hello@deeplens.space

DEEPLENS