Not just a person. 👀 What if your model could segment the face, hair, clothes, pants, shoes, and more separately?
I used the human body & accessories dataset on the @ultralytics Platform to build this demo.
More info 👇
#Fashion#MachineLearning#Retail
this is pure treasure: 20 open-source JEV combos you can plug into real products.
JEV gives you 3 primitives: Choice, Score and Noul. it makes the judgment, code enforces the rules, and another tool executes the work.
ROUTE THE WORK
1. JEV × Hermes Agent
route decisions into autonomous agent workflows.
https://t.co/D6udyyWoDq
2. JEV × LangGraph
use JEV as a decision layer inside graph-based agent flows.
https://t.co/mTmgIE2ZzB
3. JEV × MCP SDK
let JEV decide which tool or action should run next.
https://t.co/410qS3HBY4
4. JEV × Temporal
use JEV decisions inside durable workflows.
https://t.co/KiBM5twlJw
5. JEV × E2B
combine narrow decisions with sandboxed code execution.
https://t.co/hATdEoHVQq
FIND THE EVIDENCE
6. JEV × Meilisearch
retrieve relevant context before making a decision.
https://t.co/pCc9iqMkAG
7. JEV × Qdrant
score decisions against vector-retrieved evidence.
https://t.co/mE99OI4drc
8. JEV × Docling
turn documents into structured context for JEV.
https://t.co/tuMHzAiYtU
9. JEV × Graphiti
use graph memory to give JEV better context.
https://t.co/RKvtaaeZus
10. JEV × LlamaIndex
connect retrieval pipelines directly to the decision layer.
https://t.co/p8OakXtEKD
VERIFY THE RESULT
11. JEV × Langfuse
trace and inspect decisions in production.
https://t.co/myfxAVFrLl
12. JEV × Promptfoo
test JEV decisions against evals and failure cases.
https://t.co/sGCAeJKU0h
13. JEV × Semgrep
judge findings, then let code enforce security rules.
https://t.co/INgdNlpgOv
14. JEV × Sentry
route errors into the right recovery path.
https://t.co/WydOSioGsp
15. JEV × Playwright
verify browser actions before moving to the next step.
https://t.co/v1FiqiESCa
OPERATE THE PRODUCT
16. JEV × GitHub MCP
decide when to inspect repos, issues or code changes.
https://t.co/sePKfEIUAa
17. JEV × PostHog
use product data to score what should happen next.
https://t.co/C3mR5k3yQP
18. JEV × Resend
decide when and what email action should run.
https://t.co/geO9AshXq1
19. JEV × Stripe
put a decision layer in front of payment workflows.
https://t.co/fl0VkuHP1j
20. JEV × Cloudflare Workers AI
run lightweight decision flows close to the edge.
https://t.co/u2GtlzxpNM
the loop:
collect state → ask JEV a precise Choice, Score or Noul question → check confidence and policy in code → execute or escalate to a human → log the result.
these are architecture ideas, not official integrations.
save this and build a second brain for your product.
A fully open source mocap system that works with cheap webcams:
The FreeMoCap Project
A free-and-open-source, hardware-and-software-agnostic, minimal-cost, research-grade, motion capture system and platform for decentralized scientific research, education, and training:
https://t.co/GFCH6NrkGG
—-
Weekly robotics and AI insights.
Subscribe free: https://t.co/9Nm01QUcw3
MusicAgent: An AI Agent for Music Understanding and Generation with Large Language Models
paper page: https://t.co/7Np4dcjL9Z
AI-empowered music processing is a diverse field that encompasses dozens of tasks, ranging from generation tasks (e.g., timbre synthesis) to comprehension tasks (e.g., music classification). For developers and amateurs, it is very difficult to grasp all of these task to satisfy their requirements in music processing, especially considering the huge differences in the representations of music data and the model applicability across platforms among various tasks. Consequently, it is necessary to build a system to organize and integrate these tasks, and thus help practitioners to automatically analyze their demand and call suitable tools as solutions to fulfill their requirements. Inspired by the recent success of large language models (LLMs) in task automation, we develop a system, named MusicAgent, which integrates numerous music-related tools and an autonomous workflow to address user requirements. More specifically, we build 1) toolset that collects tools from diverse sources, including Hugging Face, GitHub, and Web API, etc. 2) an autonomous workflow empowered by LLMs (e.g., ChatGPT) to organize these tools and automatically decompose user requests into multiple sub-tasks and invoke corresponding music tools. The primary goal of this system is to free users from the intricacies of AI-music tools, enabling them to concentrate on the creative aspect. By granting users the freedom to effortlessly combine tools, the system offers a seamless and enriching music experience.
A developer who goes by calesthio on GitHub open sourced a full video production studio.
It is called OpenMontage. You describe what you want in plain English. The AI handles research, scripting, asset generation, voiceover, music, editing, and rendering.
Not a 4-second clip generator. A complete production system that outputs finished, edited, multi-scene videos.
12 pipelines. 52 tools. 400+ agent skills. Hit number 1 on GitHub Trending on launch day.
→ Explainers, trailers, documentaries, talking heads, tutorials, shorts, podcasts, animations, product launches
→ 14 video providers including Kling, Runway Gen-4, Google Veo 3
→ Free footage from https://t.co/0tgBSq6sN9, NASA, Wikimedia, Pexels
→ 4 TTS engines including Piper for fully offline narration
→ Music generation through Suno and ElevenLabs
→ Post-production: FFmpeg, color grading, upscaling, subtitles
Here is the wildest part:
A freelance video producer charges $2,000 to $10,000 per video. A production agency charges $5,000 to $50,000.
OpenMontage produced a cinematic sci-fi trailer for approximately $3. A polished explainer for $0.15.
$0.15. Not per month. Per video.
For documentaries, the AI builds a CLIP-searchable corpus from https://t.co/0tgBSq6sN9 and NASA. Retrieves actual motion clips. Edits them into a timeline. Not animated stills. Real edited footage.
Works with Claude Code, Cursor, Copilot, Windsurf, and Codex. AGPLv3.
This is the kind of system production studios pay hundreds of thousands for.
It runs inside your code editor. Open source. $0.15 per video.
1/ We introduce TrackEverything: a 3D point tracker that tracks all points across all frames of long videos (1000+ frames).
Our key idea is to tie computation to unique 3D scene content rather than redundant 2D pixels in a video.
https://t.co/NYHPTTg0Gr
STOP BURNING MONEY ON AI API CREDITS,10 FREE OPTIONS TO TRY:
1. OPENROUTER — https://t.co/MuDiNymH8Z
2. NVIDIA NIM — https://t.co/AYACPDJcg1
3. GITHUB MODELS — https://t.co/aNipRPw04a
4. TOGETHER AI — https://t.co/EKWIN6ufrL
5. CLOUDFLARE WORKERS AI — https://t.co/9XtjsbieoC
6. GROQ — https://t.co/Hy0909Rj3h
7. CEREBRAS — https://t.co/6YFJYMmJDO
8. MISTRAL — https://t.co/M54yVQTD9l
9. COHERE — https://t.co/v4AM9xzEJq
10. GOOGLE AI STUDIO — https://t.co/prshURZnY1
BOOKMARK THIS AND HAVE FUN
I put a Cadence patchnet into a virtual fruit fly. This creates an actual fly matrix in your browser that the fly experiences as a solid world (observation is substrate-independent). All open-source, Github link on the page.
https://t.co/vImD21Si1x
5 SKILLS THAT CAN TURN YOUR AI AGENT INTO A TUTOR, YOUTUBER, DESIGNER AND FOUNDER’S RIGHT-HAND ASSISTANT:
1. TUTOR SKILLS — LEARN ALMOST ANYTHING
https://t.co/pZ4IjBFTvI
2. YOUTUBE SKILLS — MASTER THE FULL YOUTUBE WORKFLOW
https://t.co/XX1cfYB6QI
3. PLATFORM DESIGN — BUILD INTERFACES TAILORED TO EACH PLATFORM
https://t.co/IFmftoomla
4. AEGIS — ADD GUARDRAILS ACROSS YOUR PROJECT
https://t.co/DXx2sWYB3m
5. FOUNDER SKILLS — HANDLE THE EVERYDAY WORK OF RUNNING A STARTUP
https://t.co/nWG4bp4EE7
BOOKMARK THESE AND START BUILDING YOUR OWN AI EMPLOYEE.
NobodyWho just rebuilt the core idea behind Jev, the "System One" decision model everyone on X is talking about, in 25 lines of Python that run a tiny model locally with no API calls.
→ Loads Qwen3-0.6B locally through llama-cpp-python
→ Frames the task as multiple choice, like classifying an email as legitimate, spam or phishing
→ Reads the model's logits and turns them into calibrated probabilities, 88.5% phishing in their example
→ Hit 506 points on Hacker News, with the team openly calling it a parody of the Jev hype
Repo: https://t.co/5sIuP91ZR4
This is NOT Jev.
Open source. Runs on your laptop. Decides in ~27 ms, about 200× faster than waiting on a hosted LLM.
Here it is playing Tetris by itself 👇
https://t.co/eq4gP53o2A
Jev is a super fast video editor!
Edited this entire video in less than a minute, and here's how:
First, I transcribed raw vid locally with word timestamps, cut it into 16 beats, then asked Jev these seven questions about every beat:
does this line need a visual at all?
which card template? (18 to pick from, premade)
which text effect?
which transition into it?
how hard do we punch in on the face?
which sound effect?
which word gets emphasised?
Plus five about the video as a whole. Style, colour, caption treatment, energy, progress bar.
117 questions answered by jev < 1 s ~ $0.0017
Using @typesafeai via @LiteLLM