[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7DMSRJBWBM9PPN19STJTP4Y":3,"skill-same:01M7DMSRJBWBM9PPN19STJTP4Y":120},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":28,"commitSha":29,"skillDirectory":30,"forksCount":14,"repositoryHtmlUrl":31,"repositoryHomepageUrl":6,"ownerHtmlUrl":32,"ownerBlogUrl":33,"ownerDescription":34,"updatedAt":35,"install":36,"resources":49},"01M7DMSRJBWBM9PPN19STJTP4Y","ti-edgeai-generate-config",null,"Build, validate and run edgeai-gst-apps YAML configs (inputs, models, outputs, flows) on TI Edge AI boards with a C7x\u002FMMA accelerator (TDA4VM, AM68A, AM69A, AM67A, AM62A): video file, image sequence, USB webcam, CSI camera, RTSP or test-pattern input; one or several models; display with FPS overlay, saved video\u002Fimages, network stream or fakesink; mosaic layouts. Use this skill whenever the user wants to \"run my model on the TI board\", show live results on a display, use a webcam or camera with the TI demo app, record annotated video, stream detections, write an app_config \u002F edgeai-gst-apps yaml, or asks how the TI GStreamer pipeline (tiovxmultiscaler, tiovxdlpreproc, kmssink) is assembled - even if they never say \"config\".","saeedahmadicp\u002Fti-edgeai-skills","saeedahmadicp","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F48377992?v=4","01M7DMSF1D2BMTCXPHCAA69YMX","Saeed Ahmad","01M7DMSF216NNRM1Z82M0YVPFS",0,"MIT","https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: ti-edgeai-generate-config\ndescription: >\n  Build, validate and run edgeai-gst-apps YAML configs (inputs, models, outputs, flows) on TI Edge AI boards with a C7x\u002FMMA\n  accelerator (TDA4VM, AM68A, AM69A, AM67A, AM62A): video file, image sequence, USB webcam, CSI camera, RTSP or test-pattern\n  input; one or several models; display with FPS overlay, saved video\u002Fimages, network stream or fakesink; mosaic layouts.\n  Use this skill whenever the user wants to \"run my model on the TI board\", show live results on a display, use a webcam or\n  camera with the TI demo app, record annotated video, stream detections, write an app_config \u002F edgeai-gst-apps yaml, or asks\n  how the TI GStreamer pipeline (tiovxmultiscaler, tiovxdlpreproc, kmssink) is assembled - even if they never say \"config\".\nlicense: MIT\nmetadata:\n  version: 0.1.0\n  tested_on: \"TDA4VM, Edge AI SDK 11.0: image, video, classification, two-model mosaic and USB-webcam-to-file runs. Display output not confirmed\"\n---\n\n# Configs and Runs for edgeai-gst-apps\n\nWhen this skill is active, **read `references\u002Fconfig-schema.md` before writing a config by hand** and run `scripts\u002Fvalidate_config.py`\nbefore touching the board. Prerequisites and shared rules (SoC key, `[H, W]` order, display ownership): `ti-edgeai-dev`. The model\nfolder must already exist on the board (`ti-edgeai-import-model`).\n\n## Principles\n1. **Minimal first.** Build the smallest working config (one input, one model, one output) and add mosaics, overlays or extra flows\n   only when asked.\n2. **Finite input for the first run** (image sequence or video file, file output): it ends by itself and can be checked.\n3. **Look at the output.** \"exit 0\" is not a result; open a frame and check boxes, aspect and labels.\n4. **Say what ran.** Report the input kind, config, log evidence, and what was not confirmed.\n\n## Workflow\n\n1. **Collect requirements** (infer from the request, ask only for the unknowns): input kind and its *real* mode, model folder(s),\n   output kind, display size, how long to run.\n   - Camera: discover it, do not assume a node number: `BOARD=root@\u003Cboard-ip> scripts\u002Fdiscover_cameras.sh` (capture nodes, card names,\n     formats and sizes; codec nodes skipped) and choose a mode the camera really offers. Prefer the mode whose aspect matches the model input (16:9 -> 640x360 for a 640x352 model).\n     Codec nodes (hardware decoder\u002Fencoder) are not cameras; the discovery script skips them. YUYV-only webcams need `format: auto`\n     (not `jpeg`).\n   - Output size must be \u003C= input size (the app exits otherwise).\n2. **Generate**: `python3 scripts\u002Fgenerate_config.py --input usb --source \u002Fdev\u002Fvideo\u003CN> --width 640 --height 360 --format auto\n   --model \u002Fopt\u002Fmodel_zoo\u002F\u003Cmodel> --viz-threshold 0.3 --output display --perf-overlay graph -o my.yaml`\n   (several `--model` flags tile them with mosaic rectangles). Edit by hand for anything the generator does not expose;\n   `references\u002Fconfig-schema.md` lists every key.\n3. **Validate**: `python3 scripts\u002Fvalidate_config.py my.yaml --board root@\u003Cboard-ip>` (schema, name references, size\u002Fmosaic rules, and\n   a read-only SSH check that each model folder has param.yaml, dataset.yaml and a compiled net.bin). Fix every ERROR.\n4. **Install and run**, look at the output, report what you saw.\n\n## Running\n\nFinite input, saved output (the safe first test; `GST_SOC` is the SoC key from `ti-edgeai-dev\u002Freferences\u002Fplatforms.md`):\n```bash\nscp my.yaml root@\u003Cboard-ip>:\u002Fopt\u002Fedgeai-gst-apps\u002Fconfigs\u002F\nssh root@\u003Cboard-ip> 'cd \u002Fopt\u002Fedgeai-gst-apps\u002Fapps_python && export SOC=\u003CGST_SOC> TERM=xterm &&\n   timeout 120 .\u002Fapp_edgeai.py ..\u002Fconfigs\u002Fmy.yaml > \u002Ftmp\u002Frun.log 2>&1; echo exit=$?;\n   grep -aE \"VX_ZONE_ERROR\\]|rror|Fail\" \u002Ftmp\u002Frun.log | grep -v -e curses -e setupterm'\n```\nSuccess: `exit=0`, `Offloaded Nodes - N, Total Nodes - N`, output files present (`ls \u002Fopt\u002Fedgeai-test-data\u002Foutput`). Then `scp` a frame\nback and look at it. Delete test outputs afterwards (small disk).\n\nCamera to saved frames: the same with `timeout -s INT 25` (cameras never finish; exit 124 is normal).\n\nLive on a display. This changes the board's system state (the stock GUI is stopped and a transient unit is started), so:\n1. **Get explicit authorization first.** Say exactly what you will do and what the user will lose on screen, and wait for a yes.\n   Do not stop `edgeai-init` on inference from an earlier request such as \"show it on the display\". The user must also be present\n   to watch.\n2. **Record the original state** and keep it in your report: `systemctl is-active edgeai-init; systemctl is-enabled edgeai-init`\n   (and whether `edgeai-demo` already exists).\n3. Run (only after the yes):\n```bash\nssh root@\u003Cboard-ip> 'systemctl stop edgeai-init'     # the stock GUI owns the screen\nssh root@\u003Cboard-ip> 'systemd-run --unit=edgeai-demo --setenv=SOC=\u003CGST_SOC> --setenv=TERM=xterm \\\n   --setenv=PYTHONPATH=\u002Fusr\u002Flib\u002Fpython3.12\u002Fsite-packages --working-directory=\u002Fopt\u002Fedgeai-gst-apps\u002Fapps_python \\\n   script -qfc \".\u002Fapp_edgeai.py ..\u002Fconfigs\u002Fmy_display.yaml\" \u002Fdev\u002Fnull'\nssh root@\u003Cboard-ip> 'systemctl is-active edgeai-demo'      # confirm it is alive\n```\n4. **Restore on every exit path** (done, failed, or the user asks to stop): `systemctl stop edgeai-demo; systemctl start edgeai-init`\n   (only if it was active before), then check `systemctl is-active edgeai-init` and report the state you left the board in.\nWhy a unit: the status screen needs a tty (`script` provides one) and `nohup &` children of an ssh command are killed when it ends.\nThe same consent rule covers firmware updates, reboots and USB\u002Fdriver resets (`ti-edgeai-dev`, critical rule 11).\n\n**Display status: not confirmed.** With the stock GUI running, a display-sink run showed nothing on screen; after stopping it the\napp ran as a unit without errors, but a picture was never confirmed. Report display runs as unconfirmed until the user says they\nsee video; if the screen stays blank check `\u002Fsys\u002Fclass\u002Fdrm\u002Fcard*-*\u002Fstatus` and the `connector:` option\n(`references\u002Fcamera-and-display.md`).\n\n## Pipeline sanity\nWith the right `SOC` the app builds `tiovxdlcolorconvert ! tiovxmultiscaler -> {full-res branch, tiovxdlpreproc branch}`. Print it:\n`grep -a \"multifilesrc\\|v4l2src\\|tiovx\" \u002Ftmp\u002Frun.log | cut -c1-300`. `videoscale ! videoconvert` means the ARM-only path (missing `SOC`).\nDetails: `references\u002Fpipeline-anatomy.md`.\n\n## Reference Documents\n| File | Use when |\n|---|---|\n| [references\u002Fconfig-schema.md](references\u002Fconfig-schema.md) | every config key and rule |\n| [references\u002Fpipeline-anatomy.md](references\u002Fpipeline-anatomy.md) | elements, tracers, OpTIFlow, C++ app |\n| [references\u002Fcamera-and-display.md](references\u002Fcamera-and-display.md) | device discovery, formats, display ownership, USB failure modes |\n\n## Files\n- `scripts\u002Fgenerate_config.py`, `scripts\u002Fvalidate_config.py`: PyYAML-serialized output; validator checks types and ranges (unit-tested).\n- `scripts\u002Fdiscover_cameras.sh`: read-only V4L2 capture-device discovery (unit-tested with a simulated v4l2-ctl; not run on a board).\n- `assets\u002Fexample_{image,video,classification,mosaic,webcam,webcam_display}.yaml`: starting points using TI zoo models (adapt model paths).\n","# Configs and Runs for edgeai-gst-apps\n\nWhen this skill is active, **read `references\u002Fconfig-schema.md` before writing a config by hand** and run `scripts\u002Fvalidate_config.py`\nbefore touching the board. Prerequisites and shared rules (SoC key, `[H, W]` order, display ownership): `ti-edgeai-dev`. The model\nfolder must already exist on the board (`ti-edgeai-import-model`).\n\n## Principles\n1. **Minimal first.** Build the smallest working config (one input, one model, one output) and add mosaics, overlays or extra flows\n   only when asked.\n2. **Finite input for the first run** (image sequence or video file, file output): it ends by itself and can be checked.\n3. **Look at the output.** \"exit 0\" is not a result; open a frame and check boxes, aspect and labels.\n4. **Say what ran.** Report the input kind, config, log evidence, and what was not confirmed.\n\n## Workflow\n\n1. **Collect requirements** (infer from the request, ask only for the unknowns): input kind and its *real* mode, model folder(s),\n   output kind, display size, how long to run.\n   - Camera: discover it, do not assume a node number: `BOARD=root@\u003Cboard-ip> scripts\u002Fdiscover_cameras.sh` (capture nodes, card names,\n     formats and sizes; codec nodes skipped) and choose a mode the camera really offers. Prefer the mode whose aspect matches the model input (16:9 -> 640x360 for a 640x352 model).\n     Codec nodes (hardware decoder\u002Fencoder) are not cameras; the discovery script skips them. YUYV-only webcams need `format: auto`\n     (not `jpeg`).\n   - Output size must be \u003C= input size (the app exits otherwise).\n2. **Generate**: `python3 scripts\u002Fgenerate_config.py --input usb --source \u002Fdev\u002Fvideo\u003CN> --width 640 --height 360 --format auto\n   --model \u002Fopt\u002Fmodel_zoo\u002F\u003Cmodel> --viz-threshold 0.3 --output display --perf-overlay graph -o my.yaml`\n   (several `--model` flags tile them with mosaic rectangles). Edit by hand for anything the generator does not expose;\n   `references\u002Fconfig-schema.md` lists every key.\n3. **Validate**: `python3 scripts\u002Fvalidate_config.py my.yaml --board root@\u003Cboard-ip>` (schema, name references, size\u002Fmosaic rules, and\n   a read-only SSH check that each model folder has param.yaml, dataset.yaml and a compiled net.bin). Fix every ERROR.\n4. **Install and run**, look at the output, report what you saw.\n\n## Running\n\nFinite input, saved output (the safe first test; `GST_SOC` is the SoC key from `ti-edgeai-dev\u002Freferences\u002Fplatforms.md`):\n```bash\nscp my.yaml root@\u003Cboard-ip>:\u002Fopt\u002Fedgeai-gst-apps\u002Fconfigs\u002F\nssh root@\u003Cboard-ip> 'cd \u002Fopt\u002Fedgeai-gst-apps\u002Fapps_python && export SOC=\u003CGST_SOC> TERM=xterm &&\n   timeout 120 .\u002Fapp_edgeai.py ..\u002Fconfigs\u002Fmy.yaml > \u002Ftmp\u002Frun.log 2>&1; echo exit=$?;\n   grep -aE \"VX_ZONE_ERROR\\]|rror|Fail\" \u002Ftmp\u002Frun.log | grep -v -e curses -e setupterm'\n```\nSuccess: `exit=0`, `Offloaded Nodes - N, Total Nodes - N`, output files present (`ls \u002Fopt\u002Fedgeai-test-data\u002Foutput`). Then `scp` a frame\nback and look at it. Delete test outputs afterwards (small disk).\n\nCamera to saved frames: the same with `timeout -s INT 25` (cameras never finish; exit 124 is normal).\n\nLive on a display. This changes the board's system state (the stock GUI is stopped and a transient unit is started), so:\n1. **Get explicit authorization first.** Say exactly what you will do and what the user will lose on screen, and wait for a yes.\n   Do not stop `edgeai-init` on inference from an earlier request such as \"show it on the display\". The user must also be present\n   to watch.\n2. **Record the original state** and keep it in your report: `systemctl is-active edgeai-init; systemctl is-enabled edgeai-init`\n   (and whether `edgeai-demo` already exists).\n3. Run (only after the yes):\n```bash\nssh root@\u003Cboard-ip> 'systemctl stop edgeai-init'     # the stock GUI owns the screen\nssh root@\u003Cboard-ip> 'systemd-run --unit=edgeai-demo --setenv=SOC=\u003CGST_SOC> --setenv=TERM=xterm \\\n   --setenv=PYTHONPATH=\u002Fusr\u002Flib\u002Fpython3.12\u002Fsite-packages --working-directory=\u002Fopt\u002Fedgeai-gst-apps\u002Fapps_python \\\n   script -qfc \".\u002Fapp_edgeai.py ..\u002Fconfigs\u002Fmy_display.yaml\" \u002Fdev\u002Fnull'\nssh root@\u003Cboard-ip> 'systemctl is-active edgeai-demo'      # confirm it is alive\n```\n4. **Restore on every exit path** (done, failed, or the user asks to stop): `systemctl stop edgeai-demo; systemctl start edgeai-init`\n   (only if it was active before), then check `systemctl is-active edgeai-init` and report the state you left the board in.\nWhy a unit: the status screen needs a tty (`script` provides one) and `nohup &` children of an ssh command are killed when it ends.\nThe same consent rule covers firmware updates, reboots and USB\u002Fdriver resets (`ti-edgeai-dev`, critical rule 11).\n\n**Display status: not confirmed.** With the stock GUI running, a display-sink run showed nothing on screen; after stopping it the\napp ran as a unit without errors, but a picture was never confirmed. Report display runs as unconfirmed until the user says they\nsee video; if the screen stays blank check `\u002Fsys\u002Fclass\u002Fdrm\u002Fcard*-*\u002Fstatus` and the `connector:` option\n(`references\u002Fcamera-and-display.md`).\n\n## Pipeline sanity\nWith the right `SOC` the app builds `tiovxdlcolorconvert ! tiovxmultiscaler -> {full-res branch, tiovxdlpreproc branch}`. Print it:\n`grep -a \"multifilesrc\\|v4l2src\\|tiovx\" \u002Ftmp\u002Frun.log | cut -c1-300`. `videoscale ! videoconvert` means the ARM-only path (missing `SOC`).\nDetails: `references\u002Fpipeline-anatomy.md`.\n\n## Reference Documents\n| File | Use when |\n|---|---|\n| [references\u002Fconfig-schema.md](references\u002Fconfig-schema.md) | every config key and rule |\n| [references\u002Fpipeline-anatomy.md](references\u002Fpipeline-anatomy.md) | elements, tracers, OpTIFlow, C++ app |\n| [references\u002Fcamera-and-display.md](references\u002Fcamera-and-display.md) | device discovery, formats, display ownership, USB failure modes |\n\n## Files\n- `scripts\u002Fgenerate_config.py`, `scripts\u002Fvalidate_config.py`: PyYAML-serialized output; validator checks types and ranges (unit-tested).\n- `scripts\u002Fdiscover_cameras.sh`: read-only V4L2 capture-device discovery (unit-tested with a simulated v4l2-ctl; not run on a board).\n- `assets\u002Fexample_{image,video,classification,mosaic,webcam,webcam_display}.yaml`: starting points using TI zoo models (adapt model paths).\n",{"name":5,"license":15,"metadata":24,"description":27},{"version":25,"tested_on":26},"0.1.0","TDA4VM, Edge AI SDK 11.0: image, video, classification, two-model mosaic and USB-webcam-to-file runs. Display output not confirmed","Build, validate and run edgeai-gst-apps YAML configs (inputs, models, outputs, flows) on TI Edge AI boards with a C7x\u002FMMA accelerator (TDA4VM, AM68A, AM69A, AM67A, AM62A): video file, image sequence, USB webcam, CSI camera, RTSP or test-pattern input; one or several models; display with FPS overlay, saved video\u002Fimages, network stream or fakesink; mosaic layouts. Use this skill whenever the user wants to \"run my model on the TI board\", show live results on a display, use a webcam or camera with the TI demo app, record annotated video, stream detections, write an app_config \u002F edgeai-gst-apps yaml, or asks how the TI GStreamer pipeline (tiovxmultiscaler, tiovxdlpreproc, kmssink) is assembled - even if they never say \"config\".\n","main","a8e4c792de4f8fcdf7f4889d1ebac0f7ec1dd1d0","skills\u002Fti-edgeai-generate-config","https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills","https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp","https:\u002F\u002Fsaeedahmad.me","AI Research Engineer @ IKLab Inc. | Machine Learning Engineer","2026-10-02 20:30:36",{"source":8,"skillName":5,"commands":37,"prompt":47,"zipAvailable":48},[38,41,44],{"tool":39,"command":40},"npx","npx skills add saeedahmadicp\u002Fti-edgeai-skills --skill ti-edgeai-generate-config",{"tool":42,"command":43},"bunx","bunx skills add saeedahmadicp\u002Fti-edgeai-skills --skill ti-edgeai-generate-config",{"tool":45,"command":46},"pnpm","pnpm dlx skills add saeedahmadicp\u002Fti-edgeai-skills --skill ti-edgeai-generate-config","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002FSKILL.md",true,[50,56,62,66,70,74,78,82,86,90,94,98,104,108,112,116],{"path":51,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":54,"sourceHtmlUrl":55},"agents\u002Fopenai.yaml","REFERENCE","参考文档",412,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Fagents\u002Fopenai.yaml",{"path":57,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":60,"sourceHtmlUrl":61},"assets\u002Fexample_classification.yaml","ASSET","资产文件",498,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Fassets\u002Fexample_classification.yaml",{"path":63,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":64,"sourceHtmlUrl":65},"assets\u002Fexample_image.yaml",524,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Fassets\u002Fexample_image.yaml",{"path":67,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":68,"sourceHtmlUrl":69},"assets\u002Fexample_mosaic.yaml",724,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Fassets\u002Fexample_mosaic.yaml",{"path":71,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":72,"sourceHtmlUrl":73},"assets\u002Fexample_video.yaml",537,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Fassets\u002Fexample_video.yaml",{"path":75,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":76,"sourceHtmlUrl":77},"assets\u002Fexample_webcam_display.yaml",588,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Fassets\u002Fexample_webcam_display.yaml",{"path":79,"resourceType":58,"resourceTypeDesc":59,"sizeBytes":80,"sourceHtmlUrl":81},"assets\u002Fexample_webcam.yaml",584,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Fassets\u002Fexample_webcam.yaml",{"path":83,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":84,"sourceHtmlUrl":85},"evals\u002Fevals.json",4871,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Fevals\u002Fevals.json",{"path":87,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":88,"sourceHtmlUrl":89},"references\u002Fcamera-and-display.md",1989,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Freferences\u002Fcamera-and-display.md",{"path":91,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":92,"sourceHtmlUrl":93},"references\u002Fconfig-schema.md",2726,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Freferences\u002Fconfig-schema.md",{"path":95,"resourceType":52,"resourceTypeDesc":53,"sizeBytes":96,"sourceHtmlUrl":97},"references\u002Fpipeline-anatomy.md",1533,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Freferences\u002Fpipeline-anatomy.md",{"path":99,"resourceType":100,"resourceTypeDesc":101,"sizeBytes":102,"sourceHtmlUrl":103},"scripts\u002Fdiscover_cameras.sh","SCRIPT","脚本",1127,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Fscripts\u002Fdiscover_cameras.sh",{"path":105,"resourceType":100,"resourceTypeDesc":101,"sizeBytes":106,"sourceHtmlUrl":107},"scripts\u002Fgenerate_config.py",7348,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Fscripts\u002Fgenerate_config.py",{"path":109,"resourceType":100,"resourceTypeDesc":101,"sizeBytes":110,"sourceHtmlUrl":111},"scripts\u002Fvalidate_config.py",9762,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Fscripts\u002Fvalidate_config.py",{"path":113,"resourceType":100,"resourceTypeDesc":101,"sizeBytes":114,"sourceHtmlUrl":115},"tests\u002Ftest_config_tools.py",8940,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Ftests\u002Ftest_config_tools.py",{"path":117,"resourceType":100,"resourceTypeDesc":101,"sizeBytes":118,"sourceHtmlUrl":119},"tests\u002Ftest_discover_cameras.py",2110,"https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-generate-config\u002Ftests\u002Ftest_discover_cameras.py",[121,128,135,142,149],{"skillNo":122,"name":123,"nameZh":6,"description":124,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":125,"categories":126,"keywords":127,"publishedAt":19},"01M7DMTWHET9G0KBYKTDDKKTFA","other-name","x","https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fbad-skill\u002FSKILL.md",[],[],{"skillNo":129,"name":130,"nameZh":6,"description":131,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":132,"categories":133,"keywords":134,"publishedAt":19},"01M7DMTKKJ4KY9WB99W3G9VGG6","ti-edgeai-train-model","Plan and run training\u002Ffine-tuning so a vision model (detector, classifier, segmenter) deploys well on TI Edge AI processors with a C7x\u002FMMA accelerator (TDA4VM, AM68A, AM69A, AM67A, AM62A): pick a TIDL-friendly architecture and input size, choose TI's toolchain (edgeai-modelmaker, edgeai-yolox, edgeai-mmdetection, edgeai-torchvision, plain PyTorch), build datasets and leakage-safe splits, train with quantization-friendly settings, and evaluate with the same preprocessing the board uses before handing the checkpoint to ti-edgeai-import-model. Use this skill whenever the user wants to train, fine-tune, label, split or evaluate a model \"for the TI board \u002F edge NPU \u002F C7x\", asks which architecture or input size to use, why their detector scores near zero, how to make a model quantization-friendly, or how to prepare labels so the model is learnable - even before any deployment question comes up.","https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-train-model\u002FSKILL.md",[],[],{"skillNo":136,"name":137,"nameZh":6,"description":138,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":139,"categories":140,"keywords":141,"publishedAt":19},"01M7DMTBWZYHZ01VMT94A917G3","ti-edgeai-profile-pipeline","Measure and improve inference speed on TI Edge AI boards with a C7x\u002FMMA accelerator (TDA4VM, AM68A, AM69A, AM67A, AM62A): model-only latency, per-layer C7x cycle hotspots, per-element GStreamer latency and FPS of the edgeai-gst-apps pipeline, system load, and the levers that trade accuracy for speed (input size, 16-bit layers, frame rate, output size). Use this skill whenever the user asks how fast a model runs on the TI board, wants FPS, latency, \"why is it slow\", per-layer timings, which stage is the bottleneck, whether the pipeline can keep 30 fps, or how much a precision\u002Fsize change costs - even if they only say \"benchmark it on the TI device\".","https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-profile-pipeline\u002FSKILL.md",[],[],{"skillNo":143,"name":144,"nameZh":6,"description":145,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":146,"categories":147,"keywords":148,"publishedAt":19},"01M7DMT0YPKXSV6CQFVCE8ZTVD","ti-edgeai-import-model","Get a trained neural network that exports to a static-shape ONNX (a classifier, a segmentation model, or a detector with a TIDL meta-architecture such as YOLOX\u002FSSD\u002FYOLOv5-style) running on a TI Edge AI board's C7x\u002FMMA accelerator (TDA4VM, AM68A, AM69A, AM67A, AM62A): preflight the ONNX, fold input normalisation, compile and int8-calibrate with edgeai-tidl-tools in Docker (tools tag matched to the board SDK), verify offload and accuracy in host emulation, check board-vs-host outputs, package model\u002F artifacts\u002F param.yaml dataset.yaml, deploy to \u002Fopt\u002Fmodel_zoo and smoke-test with edgeai-gst-apps. Use this skill whenever the user wants to deploy, convert, compile, quantize, port or \"put my model on\" a TI Jacinto\u002FAM6xA board, mentions TIDL compilation, artifacts, tidl_net.bin, meta_arch_type, a prototxt, calibration images, int8 accuracy loss on the TI device, or asks why a compiled model fails on the board - for any architecture within those contracts, and even if they only say \"deploy to the TI device\".","https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-import-model\u002FSKILL.md",[],[],{"skillNo":150,"name":151,"nameZh":6,"description":152,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":153,"categories":154,"keywords":155,"publishedAt":19},"01M7DMSGV8BMKPGW3PXEPJRFZA","ti-edgeai-dev","Texas Instruments Edge AI SDK development for Jacinto 7 \u002F AM6xA processors with a C7x DSP + MMA deep-learning accelerator (TDA4VM, AM68A, AM69A, AM67A, AM62A): TIDL model compilation with edgeai-tidl-tools, the on-board runtime (ONNX Runtime with the TIDL execution provider), edgeai-gst-apps configs and tiovx* GStreamer plugins, model folders and param.yaml, tools-to-SDK version matching, and troubleshooting. Use this skill whenever the user mentions TIDL, C7x, MMA, TDA4VM, J721E, AM68A, AM69A, AM67A, AM62A, edgeai-tidl-tools, edgeai-gst-apps, \u002Fopt\u002Fmodel_zoo, tiovxmultiscaler, \"TI edge NPU\u002Faccelerator\", or asks how to run, deploy, debug or speed up a neural network on a TI Jacinto\u002FSitara board - even if they never say \"skill\". Start here, then hand off to ti-edgeai-import-model, ti-edgeai-generate-config, ti-edgeai-profile-pipeline or ti-edgeai-train-model.","https:\u002F\u002Fgithub.com\u002Fsaeedahmadicp\u002Fti-edgeai-skills\u002Fblob\u002Fmain\u002Fskills\u002Fti-edgeai-dev\u002FSKILL.md",[],[]]