Add GStreamer AI_ML test suite

Runner/suites/Multimedia/GSTreamer/AI_ML/AI_ML.yaml

@@ -0,0 +1,20 @@
+metadata:
+  name: gstreamer-ai-ml
+  format: "Lava-Test Test Definition 1.0"
+  description: "Validates AI/ML inference pipelines using GStreamer (gst-launch-1.0) with Qualcomm AI RT (QAIRT) runtime components for image classification and object detection use cases."
+  os:
+    - linux
+  scope:
+    - functional
+
+params:
+  TIMEOUT: 60                    # Timeout in seconds for each pipeline (default in script is 60)
+  GST_DEBUG_LEVEL: 2             # GStreamer debug level (default in script is 2)
+  OUTPUT_VIDEO_PATH: "./video_out.mp4"  # Path for encoded output video
+
+run:
+  steps:
+    - REPO_PATH=$PWD
+    - cd Runner/suites/Multimedia/GSTreamer/AI_ML/
+    - ./run.sh --timeout "${TIMEOUT}" --gstdebug "${GST_DEBUG_LEVEL}" --output-video "${OUTPUT_VIDEO_PATH}" || true
+    - $REPO_PATH/Runner/utils/send-to-lava.sh AI_ML.res || true

Runner/suites/Multimedia/GSTreamer/AI_ML/Readme.md

@@ -0,0 +1,192 @@
+# AI_ML — Runner Test
+
+This directory contains the **AI_ML** validation test for Qualcomm Linux Testkit runners.
+
+It validates **AI/ML inference pipelines** using **GStreamer (`gst-launch-1.0`)** with Qualcomm AI RT (QAIRT) runtime components for computer vision use cases:
+- **Image Classification**  
+Streams a video (`video.mp4`) through **Inception‑v3** TensorFlow‑Lite quantized model.  
+  The frames are decoded, and the top‑5 class scores are over‑laid on‑frame. The pipeline re‑encodes the video and writes the result to the path supplied via `--output-video`.
+
+- **Object Detection** \
+Streams the video (`video.mp4`) through **YoloX** TensorFlow‑Lite quantized model. After decoding, each frame is sent to model and detected bounding boxes are composited onto the video and displayed full‑screen on Wayland socket.
+
+The script is designed to be **CI/LAVA-friendly**:
+- Writes **PASS/FAIL** into `AI_ML.res`
+- Always **exits 0** (even on FAIL) to avoid terminating LAVA jobs early
+- Logs detailed output for each pipeline test
+- Downloads required AI models automatically when connected to WiFi
+
+---
+
+## Location in repo
+
+Expected path:
+
+```
+Runner/suites/Multimedia/GSTreamer/AI_ML/run.sh
+```
+
+Required shared utils (sourced from `Runner/utils` via `init_env`):
+- `functestlib.sh`
+- `lib_gstreamer.sh`
+
+---
+
+## What this test does
+
+At a high level, the test:
+
+1. Finds and sources `init_env`
+2. Sources `$TOOLS/functestlib.sh` and `$TOOLS/lib_gstreamer.sh`
+3. Sets Wayland environment variables for proper display
+4. Connects to Ethernet network to download required AI models
+5. Defines and validates **2 different AI pipeline tests** covering:
+   - Image Classification (Inception-v3)
+   - Object Detection (YoloX)
+6. For each pipeline:
+   - Checks if all required GStreamer elements are available
+   - Runs the pipeline with a timeout
+   - Validates successful execution based on GStreamer state changes
+   - Logs detailed output for diagnostics
+7. Generates a comprehensive report with pass/fail status for each pipeline
+8. Produces a final PASS/FAIL result for the entire suite
+
+---
+
+## PASS / FAIL criteria
+
+### PASS
+- **All pipelines** successfully run to PLAYING state without critical errors
+- Output file contains `AI_ML PASS`
+- Exit code 0
+
+### FAIL
+- **Any pipeline** fails to reach PLAYING state or encounters critical errors
+- Output file contains `AI_ML FAIL`
+- Exit code 1
+
+**Note:** The test always exits `0` even for FAIL. The `.res` file is the source of truth.
+
+---
+
+## Logs and artifacts
+
+By default, logs are written relative to the script working directory:
+
+```
+./AI_ML.res
+./logs/
+  object_detection_console.log
+  object_detection_gst_debug.log
+  image_classification_console.log
+  image_classification_gst_debug.log
+```
+
+Each pipeline log contains detailed GStreamer debug output that can be examined for failures.
+
+---
+
+## Dependencies
+
+### Required
+- `gst-launch-1.0`
+- `gst-inspect-1.0`
+- `curl`, `unzip` (for model downloads)
+- Internet connectivity (to download models)
+- Proper QAIRT runtime installation (QNN)
+
+### Required AI Models (downloaded automatically)
+- `inception_v3_quantized.tflite`
+- `yolox_quantized.tflite`
+- `video.mp4`
+- Labels from `labels.zip` (containing classification and detection labels)
+
+---
+
+## Pipeline validation behavior
+
+The test validates pipelines by:
+1. First checking if all required GStreamer elements are available using `gst-inspect-1.0`
+2. Running each pipeline with a timeout (default 60 seconds)
+3. Checking for the "Setting pipeline to PLAYING" message in logs
+4. Ensuring no critical errors occur after reaching PLAYING state
+5. Special handling for timeout: if pipeline reached PLAYING before timeout, it's still considered a PASS
+
+---
+
+## Usage
+
+Run:
+
+```
+./run.sh [options]
+```
+
+### Options
+
+- `--timeout <seconds>`
+  - Sets pipeline execution timeout (default: 60 seconds)
+- `--gstdebug <level>`
+  - Sets GStreamer debug level (default: 2)
+- `--output-video <path>`
+  - Path for the encoded video generated by the classification pipeline (default: ./video_out.mp4)
+
+---
+
+## Examples
+
+### 1) Basic execution with default timeout
+
+```
+./run.sh
+```
+
+### 2) Increase timeout to 120 seconds and set debug level
+
+```
+./run.sh --timeout 120 --gstdebug 3
+```
+
+### 3) Specify custom output video path
+
+```
+./run.sh --output-video /tmp/custom_video.mp4
+```
+
+---
+
+## Troubleshooting
+
+### A) "FAIL: Test directory not found"
+- Ensure the test is located in the expected directory structure within Runner
+
+### B) "Could not find init_env"
+- Verify Runner environment is properly set up
+- Check if `init_env` exists in parent directories
+
+### C) Internet connection failures
+- Confirm internet interface is available and functional
+- The script uses `ensure_network_online` which automatically connects to available networks
+- Verify internet connectivity with ping test
+
+### D) Pipeline element missing
+- Verify all required GStreamer plugins are installed
+- Check for proper QAIRT GStreamer plugin integration
+
+### E) Model download failures
+- Confirm internet connectivity
+- Verify sufficient storage space for model downloads
+
+---
+
+## Notes for CI / LAVA
+
+- The test always exits `0` even for FAIL conditions
+- Use the `.res` file for result determination:
+  - `AI_ML PASS`
+  - `AI_ML FAIL`
+- Requires Internet connectivity to download models
+- Test duration depends on platform performance and number of successful pipelines
+- Each pipeline test has its own timeout to prevent hangs
+
+---