feat: Add CSV logging and analysis tools for rollingsum plugin

- Add csv-file property to log frame statistics
- Create analyze_sma.py for automated CSV analysis with visualizations
- Add comprehensive ROLLINGSUM_GUIDE.md documentation
- Include debugging guide and threshold recommendations
- Uses uv for Python dependency management

ROLLINGSUM_GUIDE.md

@@ -0,0 +1,340 @@
+# GStreamer Rolling Sum Plugin Guide
+
+## Overview
+
+The `rollingsum` plugin analyzes video frames in real-time by tracking the mean pixel intensity of a specific column across frames. It maintains a rolling window of these values and can drop frames that deviate significantly from the rolling mean, useful for detecting and filtering unstable or anomalous frames.
+
+## How It Works
+
+1. **Column Analysis**: Extracts mean pixel intensity from a specified vertical column
+2. **Rolling Window**: Maintains a circular buffer of recent column means
+3. **Deviation Detection**: Calculates how much each frame deviates from the rolling mean
+4. **Frame Filtering**: Optionally drops frames exceeding the deviation threshold
+5. **CSV Logging**: Records all frame statistics for analysis
+
+## Plugin Properties
+
+| Property | Type | Default | Description |
+|----------|------|---------|-------------|
+| `window-size` | int | 1000 | Number of frames in rolling window (1-100000) |
+| `column-index` | int | 1 | Which vertical column to analyze (0-based) |
+| `stride` | int | 1 | Row sampling stride (1 = every row) |
+| `threshold` | double | 0.5 | Normalized deviation threshold for dropping frames (0.0-1.0) |
+| `csv-file` | string | NULL | Path to CSV file for logging (NULL = no logging) |
+
+### Understanding Normalized Deviation
+
+- **Range**: 0.0 to 1.0
+- **Calculation**: `absolute_deviation / 255.0` (for 8-bit video)
+- **Meaning**: Fraction of the full pixel range
+  - `0.001` = deviation of ~0.255 pixel values
+  - `0.01` = deviation of ~2.55 pixel values
+  - `0.1` = deviation of ~25.5 pixel values
+
+## Basic Usage
+
+### Simple Pipeline
+
+```bash
+gst-launch-1.0 idsueyesrc config-file=config.ini ! \
+  videoconvert ! \
+  video/x-raw,format=GRAY8 ! \
+  rollingsum window-size=1000 column-index=1 threshold=0.0002 ! \
+  autovideosink
+```
+
+### With CSV Logging
+
+```bash
+gst-launch-1.0 idsueyesrc config-file=config.ini exposure=0.5 ! \
+  videoconvert ! \
+  video/x-raw,format=GRAY8 ! \
+  rollingsum window-size=1000 column-index=1 threshold=0.0002 csv-file=output.csv ! \
+  fakesink
+```
+
+## Debugging
+
+### Enable Debug Output
+
+Use the `GST_DEBUG` environment variable to see detailed plugin operation:
+
+#### Windows PowerShell
+
+```powershell
+$env:GST_DEBUG="rollingsum:5"; gst-launch-1.0 [pipeline...]
+```
+
+#### Windows CMD
+
+```cmd
+set GST_DEBUG=rollingsum:5 && gst-launch-1.0 [pipeline...]
+```
+
+#### Linux/Mac
+
+```bash
+GST_DEBUG=rollingsum:5 gst-launch-1.0 [pipeline...]
+```
+
+### Debug Levels
+
+| Level | Output |
+|-------|--------|
+| `rollingsum:1` | Errors only |
+| `rollingsum:2` | Warnings |
+| `rollingsum:3` | Info messages (file open/close) |
+| `rollingsum:4` | Debug (caps negotiation) |
+| `rollingsum:5` | Log (all frame processing) |
+
+### Example Debug Output
+
+```
+0:00:04.029432200 DEBUG rollingsum gstrollingsum.c:436: Extracted column mean: 10.07
+0:00:04.032257100 DEBUG rollingsum gstrollingsum.c:466: Frame 1: mean=10.07, rolling_mean=10.07, deviation=0.00 (normalized=0.0000)
+```
+
+**Key Fields:**
+- `Frame N`: Frame number
+- `mean`: Current frame's column mean
+- `rolling_mean`: Average of last N frames (window-size)
+- `deviation`: Absolute difference
+- `normalized`: Deviation as fraction of 255
+
+### Common Debug Scenarios
+
+#### 1. Verify Plugin Loaded
+
+```bash
+gst-inspect-1.0 rollingsum
+```
+
+Should show plugin details. If not found, check `GST_PLUGIN_PATH`.
+
+#### 2. Check CSV File Creation
+
+Look for this in debug output:
+```
+INFO rollingsum: Opened CSV file: output.csv
+```
+
+#### 3. Monitor Frame Drops
+
+Look for:
+```
+DEBUG rollingsum: Dropping frame 42: deviation 0.0005 > threshold 0.0002
+```
+
+#### 4. Verify Caps Negotiation
+
+```
+DEBUG rollingsum: set_caps
+DEBUG rollingsum: Video format: GRAY8, 1224x1026
+```
+
+## CSV Analysis
+
+### CSV Format
+
+The output CSV contains:
+
+```csv
+frame,column_mean,rolling_mean,deviation,normalized_deviation,dropped
+1,10.071150,10.071150,0.000000,0.000000,0
+2,10.059454,10.065302,0.005848,0.000023,0
+...
+```
+
+### Analyze Results
+
+Use the included analysis script:
+
+```bash
+uv run analyze_sma.py output.csv
+```
+
+**Output includes:**
+- Statistical summary (min/max/mean/std)
+- Threshold recommendations based on percentiles
+- Standard deviation-based suggestions
+- Visualization plots (`output_analysis.png`)
+
+### Interpreting Results
+
+The analysis provides threshold recommendations:
+
+| Percentile | Description | Use Case |
+|------------|-------------|----------|
+| 99th | Drops top 1% | Very conservative, catch only extreme outliers |
+| 95th | Drops top 5% | Conservative, good for quality control |
+| 90th | Drops top 10% | Balanced, moderate filtering |
+| 75th | Drops top 25% | Aggressive, maximum quality |
+
+## Recommended Thresholds
+
+Based on analysis of stable camera footage:
+
+### For General Use
+
+```bash
+# Conservative (1-2% frame drop)
+threshold=0.0003
+
+# Moderate (5-10% frame drop)
+threshold=0.0002
+
+# Aggressive (20-25% frame drop)
+threshold=0.0001
+```
+
+### For Specific Scenarios
+
+**High-speed acquisition** (minimal processing):
+```bash
+window-size=100 threshold=0.0005
+```
+
+**Quality-focused** (stable scenes):
+```bash
+window-size=1000 threshold=0.0001
+```
+
+**Real-time monitoring** (fast response):
+```bash
+window-size=50 threshold=0.0002
+```
+
+## Troubleshooting
+
+### No frames being dropped (threshold too high)
+
+**Symptom**: `dropped` column always 0 in CSV
+
+**Solution**:
+1. Run with CSV logging
+2. Analyze with `uv run analyze_sma.py output.csv`
+3. Use recommended threshold from 90th-99th percentile
+
+### Too many frames dropped (threshold too low)
+
+**Symptom**: Most frames have `dropped=1`, choppy video
+
+**Solution**:
+1. Increase threshold (try doubling current value)
+2. Check if column_index is appropriate
+3. Verify video is stable (not shaking/moving)
+
+### CSV file not created
+
+**Check**:
+1. File path is writable
+2. Look for "Opened CSV file" in debug output (`GST_DEBUG=rollingsum:3`)
+3. Verify csv-file property is set correctly
+
+### Column index out of range
+
+**Symptom**:
+```
+WARNING rollingsum: Column index 1000 >= width 1224, using column 0
+```
+
+**Solution**: Set `column-index` to value < video width
+
+### Inconsistent results
+
+**Possible causes**:
+1. Window size too small (< 50 frames)
+2. Sampling moving/dynamic content
+3. Column contains edge/artifact data
+
+**Solutions**:
+- Increase `window-size` to 500-1000
+- Choose different `column-index` (avoid edges)
+- Use `stride=2` or higher for faster processing
+
+## Performance Tips
+
+1. **Larger window = more stable** but slower to adapt to scene changes
+2. **Stride > 1** reduces computation but less accurate column mean
+3. **CSV logging** has minimal performance impact
+4. **Debug level 5** can produce massive logs, use only when needed
+
+## Integration Examples
+
+### Python Script Control
+
+```python
+import subprocess
+
+# Run pipeline with CSV logging
+subprocess.run([
+    'gst-launch-1.0',
+    'idsueyesrc', 'config-file=config.ini',
+    '!', 'videoconvert',
+    '!', 'video/x-raw,format=GRAY8',
+    '!', 'rollingsum',
+        'window-size=1000',
+        'column-index=1',
+        'threshold=0.0002',
+        'csv-file=output.csv',
+    '!', 'fakesink'
+])
+
+# Analyze results
+subprocess.run(['uv', 'run', 'analyze_sma.py', 'output.csv'])
+```
+
+### Adaptive Threshold
+
+Use analysis results to set optimal threshold for next run:
+
+```python
+import pandas as pd
+
+# Analyze previous run
+df = pd.read_csv('output.csv')
+recommended_threshold = df['normalized_deviation'].quantile(0.95)
+
+print(f"Recommended threshold: {recommended_threshold:.6f}")
+```
+
+## Developer Notes
+
+### Adding New Features
+
+Key files:
+- [`gst/rollingsum/gstrollingsum.c`](gst/rollingsum/gstrollingsum.c) - Main implementation
+- [`gst/rollingsum/gstrollingsum.h`](gst/rollingsum/gstrollingsum.h) - Header/structures
+- [`gst/rollingsum/CMakeLists.txt`](gst/rollingsum/CMakeLists.txt) - Build config
+
+### Rebuild After Changes
+
+```bash
+.\build.ps1  # Windows
+./build.sh   # Linux
+```
+
+### Testing
+
+```bash
+# Quick test
+gst-inspect-1.0 rollingsum
+
+# Full pipeline test with debug
+$env:GST_DEBUG="rollingsum:5"
+gst-launch-1.0 videotestsrc ! rollingsum ! fakesink
+```
+
+## References
+
+- [DESIGN_ROLLINGSUM.md](DESIGN_ROLLINGSUM.md) - Design document
+- [analyze_sma.py](analyze_sma.py) - Analysis tool
+- GStreamer documentation: https://gstreamer.freedesktop.org/documentation/
+
+## Support
+
+For issues or questions:
+1. Enable debug output (`GST_DEBUG=rollingsum:5`)
+2. Generate CSV log and analyze
+3. Check this guide's troubleshooting section
+4. Review debug output for errors/warnings

analyze_sma.py

@@ -0,0 +1,184 @@
+#!/usr/bin/env python3
+# /// script
+# dependencies = [
+#   "pandas>=2.0.0",
+#   "matplotlib>=3.7.0",
+#   "numpy>=1.24.0",
+# ]
+# ///
+
+"""
+Rolling Sum Analysis Tool
+Analyzes CSV output from the GStreamer rollingsum plugin
+Usage: uv run analyze_sma.py [csv_file]
+"""
+
+import sys
+import pandas as pd
+import matplotlib.pyplot as plt
+import numpy as np
+from pathlib import Path
+
+
+def analyze_csv(csv_file: str = "output.csv"):
+    """Analyze the rolling sum CSV data and generate insights."""
+    
+    # Read the CSV
+    try:
+        df = pd.read_csv(csv_file)
+    except FileNotFoundError:
+        print(f"Error: CSV file '{csv_file}' not found.")
+        sys.exit(1)
+    
+    print("=" * 80)
+    print(f"ROLLING SUM ANALYSIS - {csv_file}")
+    print("=" * 80)
+    print()
+    
+    # Basic statistics
+    print("DATASET OVERVIEW:")
+    print(f"  Total frames: {len(df)}")
+    print(f"  Frames dropped: {df['dropped'].sum()}")
+    print(f"  Frames kept: {(df['dropped'] == 0).sum()}")
+    print(f"  Drop rate: {df['dropped'].mean() * 100:.2f}%")
+    print()
+    
+    # Column mean statistics
+    print("COLUMN MEAN STATISTICS:")
+    print(f"  Min: {df['column_mean'].min():.6f}")
+    print(f"  Max: {df['column_mean'].max():.6f}")
+    print(f"  Range: {df['column_mean'].max() - df['column_mean'].min():.6f}")
+    print(f"  Mean: {df['column_mean'].mean():.6f}")
+    print(f"  Std Dev: {df['column_mean'].std():.6f}")
+    print()
+    
+    # Deviation statistics
+    print("DEVIATION STATISTICS:")
+    print(f"  Min deviation: {df['deviation'].min():.6f}")
+    print(f"  Max deviation: {df['deviation'].max():.6f}")
+    print(f"  Mean deviation: {df['deviation'].mean():.6f}")
+    print(f"  Std dev of deviations: {df['deviation'].std():.6f}")
+    print()
+    
+    # Normalized deviation statistics
+    print("NORMALIZED DEVIATION STATISTICS:")
+    print(f"  Min: {df['normalized_deviation'].min():.8f}")
+    print(f"  Max: {df['normalized_deviation'].max():.8f}")
+    print(f"  Mean: {df['normalized_deviation'].mean():.8f}")
+    print(f"  Median: {df['normalized_deviation'].median():.8f}")
+    print(f"  95th percentile: {df['normalized_deviation'].quantile(0.95):.8f}")
+    print(f"  99th percentile: {df['normalized_deviation'].quantile(0.99):.8f}")
+    print()
+    
+    # Threshold recommendations
+    print("THRESHOLD RECOMMENDATIONS:")
+    print("  (Based on normalized deviation percentiles)")
+    print()
+    
+    percentiles = [50, 75, 90, 95, 99]
+    for p in percentiles:
+        threshold = df['normalized_deviation'].quantile(p / 100)
+        frames_dropped = (df['normalized_deviation'] > threshold).sum()
+        drop_rate = (frames_dropped / len(df)) * 100
+        print(f"  {p}th percentile: threshold={threshold:.8f}")
+        print(f"    → Would drop {frames_dropped} frames ({drop_rate:.1f}%)")
+        print()
+    
+    # Suggest optimal thresholds based on standard deviations
+    mean_norm_dev = df['normalized_deviation'].mean()
+    std_norm_dev = df['normalized_deviation'].std()
+    
+    print("STANDARD DEVIATION-BASED THRESHOLDS:")
+    for n in [1, 2, 3]:
+        threshold = mean_norm_dev + (n * std_norm_dev)
+        frames_dropped = (df['normalized_deviation'] > threshold).sum()
+        drop_rate = (frames_dropped / len(df)) * 100
+        print(f"  Mean + {n}σ: threshold={threshold:.8f}")
+        print(f"    → Would drop {frames_dropped} frames ({drop_rate:.1f}%)")
+        print()
+    
+    # Create visualizations
+    create_plots(df, csv_file)
+    
+    print("=" * 80)
+    print("PLOTS SAVED:")
+    print(f"  - {csv_file.replace('.csv', '_analysis.png')}")
+    print("=" * 80)
+
+
+def create_plots(df: pd.DataFrame, csv_file: str):
+    """Create analysis plots."""
+    
+    fig, axes = plt.subplots(2, 2, figsize=(14, 10))
+    fig.suptitle(f'Rolling Sum Analysis - {csv_file}', fontsize=16, fontweight='bold')
+    
+    # Plot 1: Column mean over time
+    ax1 = axes[0, 0]
+    ax1.plot(df['frame'], df['column_mean'], label='Column Mean', linewidth=1)
+    ax1.plot(df['frame'], df['rolling_mean'], label='Rolling Mean', linewidth=1, alpha=0.7)
+    ax1.set_xlabel('Frame')
+    ax1.set_ylabel('Pixel Value')
+    ax1.set_title('Column Mean vs Rolling Mean')
+    ax1.legend()
+    ax1.grid(True, alpha=0.3)
+    
+    # Plot 2: Deviation over time
+    ax2 = axes[0, 1]
+    ax2.plot(df['frame'], df['deviation'], linewidth=1, color='orange')
+    ax2.axhline(y=df['deviation'].mean(), color='r', linestyle='--', 
+                label=f'Mean: {df["deviation"].mean():.4f}')
+    ax2.axhline(y=df['deviation'].quantile(0.95), color='g', linestyle='--',
+                label=f'95th: {df["deviation"].quantile(0.95):.4f}')
+    ax2.set_xlabel('Frame')
+    ax2.set_ylabel('Absolute Deviation')
+    ax2.set_title('Deviation from Rolling Mean')
+    ax2.legend()
+    ax2.grid(True, alpha=0.3)
+    
+    # Plot 3: Normalized deviation distribution
+    ax3 = axes[1, 0]
+    ax3.hist(df['normalized_deviation'], bins=50, edgecolor='black', alpha=0.7)
+    ax3.axvline(x=df['normalized_deviation'].mean(), color='r', linestyle='--',
+                label=f'Mean: {df["normalized_deviation"].mean():.6f}')
+    ax3.axvline(x=df['normalized_deviation'].median(), color='g', linestyle='--',
+                label=f'Median: {df["normalized_deviation"].median():.6f}')
+    ax3.set_xlabel('Normalized Deviation')
+    ax3.set_ylabel('Frequency')
+    ax3.set_title('Normalized Deviation Distribution')
+    ax3.legend()
+    ax3.grid(True, alpha=0.3, axis='y')
+    
+    # Plot 4: Cumulative distribution
+    ax4 = axes[1, 1]
+    sorted_norm_dev = np.sort(df['normalized_deviation'])
+    cumulative = np.arange(1, len(sorted_norm_dev) + 1) / len(sorted_norm_dev) * 100
+    ax4.plot(sorted_norm_dev, cumulative, linewidth=2)
+    
+    # Mark percentiles
+    for p in [50, 75, 90, 95, 99]:
+        threshold = df['normalized_deviation'].quantile(p / 100)
+        ax4.axvline(x=threshold, color='red', linestyle=':', alpha=0.5)
+        ax4.text(threshold, p, f'{p}th', rotation=90, va='bottom', ha='right', fontsize=8)
+    
+    ax4.set_xlabel('Normalized Deviation')
+    ax4.set_ylabel('Cumulative Percentage (%)')
+    ax4.set_title('Cumulative Distribution Function')
+    ax4.grid(True, alpha=0.3)
+    
+    plt.tight_layout()
+    
+    # Save the plot
+    output_file = csv_file.replace('.csv', '_analysis.png')
+    plt.savefig(output_file, dpi=150, bbox_inches='tight')
+    print(f"\n✓ Saved analysis plot to: {output_file}\n")
+
+
+if __name__ == "__main__":
+    csv_file = sys.argv[1] if len(sys.argv) > 1 else "output.csv"
+    
+    if not Path(csv_file).exists():
+        print(f"Error: File '{csv_file}' not found.")
+        print(f"Usage: uv run analyze_sma.py [csv_file]")
+        sys.exit(1)
+    
+    analyze_csv(csv_file)

gst/rollingsum/gstrollingsum.c

@@ -38,6 +38,7 @@
 #include "gstrollingsum.h"
 #include <string.h>
 #include <math.h>
+#include <stdio.h>
 
 enum
 {
@@ -46,6 +47,7 @@ enum
   PROP_COLUMN_INDEX,
   PROP_STRIDE,
   PROP_THRESHOLD,
+  PROP_CSV_FILENAME,
   PROP_LAST
 };
 
@@ -53,6 +55,7 @@ enum
 #define DEFAULT_PROP_COLUMN_INDEX 1
 #define DEFAULT_PROP_STRIDE 1
 #define DEFAULT_PROP_THRESHOLD 0.5
+#define DEFAULT_PROP_CSV_FILENAME NULL
 
 /* Supported video formats */
 #define SUPPORTED_CAPS \
@@ -108,6 +111,18 @@ gst_rolling_sum_dispose (GObject * object)
 
   GST_DEBUG ("dispose");
 
+  /* Close CSV file if open */
+  if (filter->csv_file) {
+    fclose (filter->csv_file);
+    filter->csv_file = NULL;
+  }
+
+  /* Free CSV filename */
+  if (filter->csv_filename) {
+    g_free (filter->csv_filename);
+    filter->csv_filename = NULL;
+  }
+
   gst_rolling_sum_reset (filter);
 
   /* chain up to the parent class */
@@ -161,6 +176,13 @@ gst_rolling_sum_class_init (GstRollingSumClass * klass)
           G_PARAM_STATIC_STRINGS | G_PARAM_READWRITE |
           GST_PARAM_MUTABLE_PLAYING));
 
+  g_object_class_install_property (gobject_class, PROP_CSV_FILENAME,
+      g_param_spec_string ("csv-file", "CSV File",
+          "Path to CSV file for logging frame data (NULL = no logging)",
+          DEFAULT_PROP_CSV_FILENAME,
+          G_PARAM_STATIC_STRINGS | G_PARAM_READWRITE |
+          GST_PARAM_MUTABLE_READY));
+
   gst_element_class_add_pad_template (gstelement_class,
       gst_static_pad_template_get (&gst_rolling_sum_sink_template));
   gst_element_class_add_pad_template (gstelement_class,
@@ -187,6 +209,7 @@ gst_rolling_sum_init (GstRollingSum * filter)
   filter->column_index = DEFAULT_PROP_COLUMN_INDEX;
   filter->stride = DEFAULT_PROP_STRIDE;
   filter->threshold = DEFAULT_PROP_THRESHOLD;
+  filter->csv_filename = NULL;
 
   filter->ring_buffer = NULL;
   filter->ring_index = 0;
@@ -194,6 +217,7 @@ gst_rolling_sum_init (GstRollingSum * filter)
   filter->rolling_mean = 0.0;
   filter->rolling_sum = 0.0;
   filter->info_set = FALSE;
+  filter->csv_file = NULL;
 
   gst_base_transform_set_in_place (GST_BASE_TRANSFORM (filter), TRUE);
 
@@ -228,6 +252,40 @@ gst_rolling_sum_set_property (GObject * object, guint prop_id,
     case PROP_THRESHOLD:
       filter->threshold = g_value_get_double (value);
       break;
+    case PROP_CSV_FILENAME:
+    {
+      const gchar *filename = g_value_get_string (value);
+      
+      /* Close old file if open */
+      if (filter->csv_file) {
+        fclose (filter->csv_file);
+        filter->csv_file = NULL;
+      }
+      
+      /* Free old filename */
+      if (filter->csv_filename) {
+        g_free (filter->csv_filename);
+        filter->csv_filename = NULL;
+      }
+      
+      /* Set new filename and open file */
+      if (filename && filename[0] != '\0') {
+        filter->csv_filename = g_strdup (filename);
+        filter->csv_file = fopen (filter->csv_filename, "w");
+        
+        if (filter->csv_file) {
+          /* Write CSV header */
+          fprintf (filter->csv_file, "frame,column_mean,rolling_mean,deviation,normalized_deviation,dropped\n");
+          fflush (filter->csv_file);
+          GST_INFO_OBJECT (filter, "Opened CSV file: %s", filter->csv_filename);
+        } else {
+          GST_ERROR_OBJECT (filter, "Failed to open CSV file: %s", filter->csv_filename);
+          g_free (filter->csv_filename);
+          filter->csv_filename = NULL;
+        }
+      }
+      break;
+    }
     default:
       G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
       break;
@@ -255,6 +313,9 @@ gst_rolling_sum_get_property (GObject * object, guint prop_id, GValue * value,
     case PROP_THRESHOLD:
       g_value_set_double (value, filter->threshold);
       break;
+    case PROP_CSV_FILENAME:
+      g_value_set_string (value, filter->csv_filename);
+      break;
     default:
       G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
       break;
@@ -367,8 +428,12 @@ gst_rolling_sum_transform_ip (GstBaseTransform * trans, GstBuffer * buf)
   gdouble frame_mean, deviation, old_value;
   gint effective_window_size;
 
+  GST_DEBUG_OBJECT (filter, "transform_ip called, frame_count=%d", filter->frame_count);
+
   /* Extract column mean from current frame */
   frame_mean = gst_rolling_sum_extract_column_mean (filter, buf);
+  
+  GST_DEBUG_OBJECT (filter, "Extracted column mean: %.2f", frame_mean);
 
   /* Store in ring buffer */
   old_value = filter->ring_buffer[filter->ring_index];
@@ -395,7 +460,7 @@ gst_rolling_sum_transform_ip (GstBaseTransform * trans, GstBuffer * buf)
   /* Normalize deviation (assuming 8-bit equivalent range) */
   gdouble normalized_deviation = deviation / 255.0;
 
-  GST_LOG_OBJECT (filter,
+  GST_DEBUG_OBJECT (filter,
       "Frame %d: mean=%.2f, rolling_mean=%.2f, deviation=%.2f (normalized=%.4f)",
       filter->frame_count, frame_mean, filter->rolling_mean, deviation,
       normalized_deviation);
@@ -404,10 +469,23 @@ gst_rolling_sum_transform_ip (GstBaseTransform * trans, GstBuffer * buf)
   filter->ring_index = (filter->ring_index + 1) % filter->window_size;
 
   /* Decision: drop or pass frame */
+  gboolean dropped = FALSE;
   if (normalized_deviation > filter->threshold) {
     GST_DEBUG_OBJECT (filter,
         "Dropping frame %d: deviation %.4f > threshold %.4f",
         filter->frame_count, normalized_deviation, filter->threshold);
+    dropped = TRUE;
+  }
+
+  /* Write to CSV if file is open */
+  if (filter->csv_file) {
+    fprintf (filter->csv_file, "%d,%.6f,%.6f,%.6f,%.6f,%d\n",
+        filter->frame_count, frame_mean, filter->rolling_mean,
+        deviation, normalized_deviation, dropped ? 1 : 0);
+    fflush (filter->csv_file);
+  }
+
+  if (dropped) {
     return GST_BASE_TRANSFORM_FLOW_DROPPED;
   }
 

gst/rollingsum/gstrollingsum.h

@@ -22,6 +22,7 @@
 
 #include <gst/base/gstbasetransform.h>
 #include <gst/video/video.h>
+#include <stdio.h>
 
 G_BEGIN_DECLS
 
@@ -54,6 +55,7 @@ struct _GstRollingSum
   gint column_index;     /* Which column to analyze (0-based) */
   gint stride;           /* Row sampling stride */
   gdouble threshold;     /* Deviation threshold for dropping frames */
+  gchar *csv_filename;   /* CSV output filename (NULL = no CSV) */
 
   /* State */
   gdouble *ring_buffer;  /* Circular buffer of column means */
@@ -61,6 +63,7 @@ struct _GstRollingSum
   gint frame_count;      /* Total frames processed */
   gdouble rolling_mean;  /* Current rolling mean */
   gdouble rolling_sum;   /* Current rolling sum for efficient mean update */
+  FILE *csv_file;        /* CSV file handle */
   
   /* Video format info */
   GstVideoInfo video_info;