# AI use cases for JEV Sees (CharlesFeng0314/JEV_sees) Repo: [CharlesFeng0314/JEV_sees](https://github.com/CharlesFeng0314/JEV_sees) | Use cases: 5 | Page: https://dedupedai.com/use-cases/charlesfeng0314-jev-sees/ JEV Sees is an open-source Python project that detects objects in images, video, and RGB-D camera feeds, tracks them across frames, and asks the JEV model a question about each one. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Ask JEV a color question about each object in a photo: https://dedupedai.com/use-cases/charlesfeng0314-jev-sees/answer-color-question-per-object-in-photo/ - Score each tracked pedestrian's accident probability in a clip: https://dedupedai.com/use-cases/charlesfeng0314-jev-sees/track-pedestrian-risk-across-traffic-clip/ - Label each tracked animal's motion state in a video: https://dedupedai.com/use-cases/charlesfeng0314-jev-sees/label-animal-motion-state-in-video/ - Grade how obstructed each package is in a photo: https://dedupedai.com/use-cases/charlesfeng0314-jev-sees/grade-package-obstruction-in-warehouse-photo/ - Ask JEV questions about objects from a live RGB-D camera: https://dedupedai.com/use-cases/charlesfeng0314-jev-sees/ask-jev-about-rgbd-camera-objects-live/ ## How to ask JEV a color question about each object in a photo with JEV Sees Repo: [CharlesFeng0314/JEV_sees](https://github.com/CharlesFeng0314/JEV_sees) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Detect objects in one image and get a per-object color answer from a fixed list of options. **Who it is for:** A developer or analyst with a single photo who needs to label an attribute, such as the color of each vehicle, and wants structured rows back. **The problem:** Answering a visual question for every object in a picture by hand is slow and hard to repeat. **The outcome:** A terminal table with one row per detected object showing the object ID, the chosen color, and a confidence value. ### Steps 1. Clone the project with `git clone https://github.com/CharlesFeng0314/JEV_sees.git`. 2. Enter the folder with `cd JEV_sees`. 3. Install it with `python -m pip install -e .`. 4. Set your key on macOS or Linux with `export TYPESAFE_API_KEY="your-key"`, or in PowerShell with `$env:TYPESAFE_API_KEY = "your-key"`. 5. Run the bundled example with `python examples/bus_color.py`. ### Prompt to paste into your agent ```text In the JEV_sees repo at , with TYPESAFE_API_KEY set, run examples/bus_color.py and report the printed rows for assets/bus.jpg. Expect a table with object_id, label, choice, and confidence columns; verify that each choice is one of the listed colors. ``` ### Caveat JEV Sees is an independent experimental project (v0.1.0) built on JEV, not an official TypeSafe AI product, and its API may change. Risk, credentials: Keep your TYPESAFE_API_KEY in your shell environment and do not paste it into committed files or shared logs. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/charlesfeng0314-jev-sees/answer-color-question-per-object-in-photo/ ## How to score each tracked pedestrian's accident probability in a clip with JEV Sees Repo: [CharlesFeng0314/JEV_sees](https://github.com/CharlesFeng0314/JEV_sees) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Get a per-frame probability for every tracked pedestrian in a short traffic video segment. **Who it is for:** A traffic safety analyst reviewing a dashcam or street camera clip who needs to see which pedestrians are judged as involved in a car accident in the current frame. **The problem:** Reviewing many frames by hand to follow each pedestrian and judge their involvement is slow and inconsistent. **The outcome:** A table of frame indexes, video times, object IDs, and probabilities for each pedestrian in the chosen segment. ### Steps 1. Clone the project with `git clone https://github.com/CharlesFeng0314/JEV_sees.git`. 2. Enter the folder with `cd JEV_sees`. 3. Install it with `python -m pip install -e .`. 4. Set your key on macOS or Linux with `export TYPESAFE_API_KEY="your-key"`, or in PowerShell with `$env:TYPESAFE_API_KEY = "your-key"`. 5. Run the bundled example with `python examples/traffic_relations.py`. 6. To test another segment, change the start and end frame numbers in your own script that calls probability(...). ### Prompt to paste into your agent ```text In the JEV_sees repo at , with TYPESAFE_API_KEY set, run examples/traffic_relations.py on assets/traffic.mp4 for frames 504 to 600. Report frame_index, object_id, and probability for each pedestrian; verify that the output describes the current frame and does not forecast future accidents. ``` ### Caveat The README says this asks about the current frame and is not a future-accident forecast, so do not use the output as a prediction. Risk, credentials: Store the JEV API key in your environment only, since each run sends clip frames to the JEV service. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/charlesfeng0314-jev-sees/track-pedestrian-risk-across-traffic-clip/ ## How to label each tracked animal's motion state in a video with JEV Sees Repo: [CharlesFeng0314/JEV_sees](https://github.com/CharlesFeng0314/JEV_sees) | Difficulty: Intermediate | Time: 30 minutes | Basis: From the README Assign stationary, walking, running, or uncertain to every tracked animal in a clip. **Who it is for:** A wildlife researcher or pet owner with a camera recording who wants a timeline of how an animal is moving. **The problem:** Watching long clips to note when an animal stops, walks, or runs takes a lot of time and is easy to get wrong. **The outcome:** A list of frames with object IDs, labels, and a motion choice with confidence for each sampled frame. ### Steps 1. Clone the project with `git clone https://github.com/CharlesFeng0314/JEV_sees.git`. 2. Enter the folder with `cd JEV_sees`. 3. Install it with `python -m pip install -e .`. 4. Set your key on macOS or Linux with `export TYPESAFE_API_KEY="your-key"`, or in PowerShell with `$env:TYPESAFE_API_KEY = "your-key"`. 5. Run the bundled example with `python examples/animal_motion.py`. ### Prompt to paste into your agent ```text In the JEV_sees repo at , with TYPESAFE_API_KEY set, run examples/animal_motion.py on assets/black_cat_walking.mp4. Report frame_index, video_time_s, object_id, and choice for the output; verify that every choice is one of stationary, walking, running, or uncertain. ``` ### Caveat The README examples use sample media, and the default sampling rate is about twice per second, so fast motion between samples may be missed. Risk, credentials: Keep your JEV API key out of shared scripts and screenshots of the terminal output. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/charlesfeng0314-jev-sees/label-animal-motion-state-in-video/ ## How to grade how obstructed each package is in a photo with JEV Sees Repo: [CharlesFeng0314/JEV_sees](https://github.com/CharlesFeng0314/JEV_sees) | Difficulty: Intermediate | Time: 45 minutes | Basis: From the README Score every detected package in a warehouse image against a rubric you define. **Who it is for:** A warehouse operations lead who reviews shelf or dock photos and needs a consistent obstruction grade for each package. **The problem:** Checking each package in many photos against the same standard by eye produces uneven results. **The outcome:** A structured export of one score per package, with the rubric text the application supplied. ### Steps 1. Clone the project with `git clone https://github.com/CharlesFeng0314/JEV_sees.git`. 2. Enter the folder with `cd JEV_sees`. 3. Install it with `python -m pip install -e .`. 4. Set your key on macOS or Linux with `export TYPESAFE_API_KEY="your-key"`, or in PowerShell with `$env:TYPESAFE_API_KEY = "your-key"`. 5. Write a short Python script that imports Sees from jev_sees and calls Sees().for_each("package").score(...) on your photo with your rubric, then sets json="output.json" to export the results. (inferred, check the README) ### Prompt to paste into your agent ```text In the JEV_sees repo at , write and run a Python script that calls Sees().for_each("package").score(...) on with the rubric 0: clear, 1: partly obstructed, 2: heavily obstructed. Expect one score per package in output.json; verify that every score is one of the three rubric levels. ``` ### Caveat The score method is part of an experimental v0.1.0 API that the README says may continue to change, so pin the version you test. Risk, credentials: Use an environment variable for TYPESAFE_API_KEY rather than hard-coding the key in the script. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/charlesfeng0314-jev-sees/grade-package-obstruction-in-warehouse-photo/ ## How to ask JEV questions about objects from a live RGB-D camera with JEV Sees Repo: [CharlesFeng0314/JEV_sees](https://github.com/CharlesFeng0314/JEV_sees) | Difficulty: Advanced | Time: 60 minutes | Basis: From the README Feed RGB and depth frames into a loop and ask JEV about detected objects with 3D positions. **Who it is for:** A robotics developer whose camera publishes RGB and depth frames and who needs answers about objects that include metric positions. **The problem:** Connecting a live depth camera to a visual question model usually needs custom glue code for object positions and per-frame answers. **The outcome:** A running camera loop that prints JEV answers for each observed object, with distances when camera intrinsics are provided. ### Steps 1. Clone the project with `git clone https://github.com/CharlesFeng0314/JEV_sees.git`. 2. Enter the folder with `cd JEV_sees`. 3. Install it with `python -m pip install -e .`. 4. Set your key on macOS or Linux with `export TYPESAFE_API_KEY="your-key"`, or in PowerShell with `$env:TYPESAFE_API_KEY = "your-key"`. 5. Write a loop that reads RGB and depth frames from your camera and calls observe(rgb, depth, intrinsics=...), then passes the returned objects to ask(...) with your question. (inferred, check the README) ### Prompt to paste into your agent ```text In the JEV_sees repo at , write a camera loop that calls observe(rgb, depth, intrinsics=...) on each frame from and then ask(...) with the question What is this object? Expect printed JEV answers per object; verify that objects include 3D positions or distances when intrinsics are supplied. ``` ### Caveat The README mentions observe and ask only briefly without full signatures, so confirm the exact call format in the source before relying on it. Risk, credentials: Keep the JEV API key in your environment and out of the camera script you share with others. Dual-use notice: only use this on software, sites and systems you own or are authorized to test. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/charlesfeng0314-jev-sees/ask-jev-about-rgbd-camera-objects-live/