Read and display CARLA object bounding boxes, then project them onto a camera image as 2D rectangles. Export an annotated PNG and box coordinates in JSON for object detection annotation work.
中文简介正在准备中。当前显示英语简介。
Gets actor and level 3D bounding boxes and projects them into a camera image as 2D boxes — the dataset-annotation workflow. Lists box geometry, draws 3D boxes in the world via debug, or captures one camera frame and writes an annotated PNG plus a JSON of 2D boxes for matching actors. Use when the user asks to "get bounding boxes", "draw boxes around the cars", "project boxes onto the camera", or "generate a detection dataset / annotations".
安装前可以查看交给 AI 代理的指令内容。
Paths.
scripts/…andreferences/…below are relative to the directory holding this SKILL.md. Your working directory is the user's project, not that directory, so prefix them with its absolute path or the command is not found.
Three things: read actors' 3D boxes, draw them in the world, or project them
into a camera image as 2D boxes with a JSON sidecar — the standard CARLA
dataset-annotation pipeline (client_bounding_boxes.py).
Progress:
- [ ] Step 1: Check prerequisites (bash scripts/check_env.sh), clear FAILs
- [ ] Step 2: For `project`, have a camera on the ego (create-sensor) + actors around
- [ ] Step 3: list / draw / project
Commands need CARLA_HOST/CARLA_PORT from scripts/env.sh.
bash scripts/check_env.sh
source scripts/env.sh
# 3D box geometry of every vehicle
python3 scripts/bounding_boxes.py list --filter vehicle.*
# draw 3D boxes around vehicles — they track moving cars, stay 30 s (default)
python3 scripts/bounding_boxes.py draw --filter vehicle.*
python3 scripts/bounding_boxes.py draw --filter vehicle.* --seconds 60
# project vehicle boxes into a camera frame -> annotated PNG + JSON
python3 scripts/bounding_boxes.py project --camera <camera_id> --filter vehicle.* --out boxes.png
Capture one frame → build the camera intrinsics K from its fov + resolution →
for each actor, take its 3D box's 8 world vertices (get_world_vertices),
transform to camera space with the camera's inverse matrix, keep those in front,
apply K → 2D points → the min/max is the 2D box. Off-frame and too-far
(--max-dist) actors are skipped. Full math in the reference.
Example 1: annotate the ego camera
User says: "put boxes on the cars in the camera view"
Spawn a camera on the ego (create-sensor), then project --camera <id> --filter vehicle.* → boxes.png with green 2D boxes + boxes.json.
Example 2: how big is that actor
User says: "what are the dimensions of vehicle 137?"
list --filter vehicle.* (find 137) → its LxWxH from the extent.
Example 3: visualise in 3D
User says: "outline all the vehicles in the world"
draw --filter vehicle.* — green 3D boxes on a rendered server.
Problem: project finds no boxes
Cause: no actors in front of / within --max-dist of the camera, or wrong filter.
Solution: raise --max-dist, widen --filter, ensure vehicles are in view.
Problem: boxes misaligned Cause: the frame and the actor poses drifted (fast motion, async). Solution: capture in synchronous mode (set-world-settings) so the frame and poses are from the same tick.
Problem: draw shows nothing
Cause: headless -nullrhi server.
Solution: use a rendered server (see debug-draw).
Note: boxes always track moving actors
draw re-stamps each frame for --seconds (default 30), so boxes ride moving
cars automatically — the command blocks for that duration.
list: box dimensions per actor.draw: 3D box overlay (rendered view).project: <out>.png (annotated) + <out>.json (2D boxes: id, type, xyxy).Detail (intrinsics, world→camera transform, level boxes, lidar-to-camera) in references/bounding-boxes.md.
还没有评价。欢迎分享你的使用感受。
无需登录即可发表。欢迎分享使用感受、提问，或说明在哪些环境下无法运行。
还没有评论。来写第一条吧。
Add a message type to CARLA's native ROS 2 interface. Create its C++ data structure, serialization code for transmission, type identity hash, registration, and tests, with Docker-based hash calculation and build checks.
中文简介正在准备中。当前显示英语简介。
Extend CARLA’s native ROS 2 interface to publish sensor data or receive vehicle commands. Covers topic implementation, registration, data connections, and verification, plus adding middleware, the communication layer used to exchange messages.
中文简介正在准备中。当前显示英语简介。
Summarize ScenarioRunner results to see which scenarios passed and which checks failed. Compare runs before and after changes, or analyze recorded driving to produce measurements and plots of vehicle distance and deviation from the lane center.
中文简介正在准备中。当前显示英语简介。
Checks a local CARLA checkout for missing API documentation, outdated references, broken links, and pages absent from navigation. Produces a grouped Markdown report without building the project or running the simulator.
中文简介正在准备中。当前显示英语简介。
Builds CARLA's ue4-dev branch on Linux, producing the Unreal Engine 4.26 editor, server modules, and an installed Python client package. Fetches required content, optionally compiles native ROS 2 support, and verifies the build against a source-built server.
中文简介正在准备中。当前显示英语简介。
Build CARLA for Unreal Engine 5.8 using CMake, the tool that configures its build process. The skill guides engine setup, optional ROS 2 support, Python API rebuilding, editor builds, and checks that the expected files were actually produced.
中文简介正在准备中。当前显示英语简介。