# CNN v2 Testing Tool

WebGPU-based browser tool for testing trained CNN v2 weights.

---

## Features

- Drag-drop PNG images and `.bin` weights (or click to browse)
- Real-time CNN inference with WebGPU compute shaders
- View modes: CNN output, original input, difference (×10)
- Adjustable blend amount and depth
- Data-driven pipeline (supports variable layer count)
- GPU timing display
- **Left Panel:** Weights info + kernel visualization (1px/weight, all layers)
- **Right Panel:** Layer activation viewer with 4-channel split + 4× zoom

---

## Requirements

- Browser with WebGPU support:
  - Chrome/Edge 113+ (enable `chrome://flags/#enable-unsafe-webgpu` if needed)
  - Safari 18+ (macOS Ventura+)
- Trained CNN v2 weights in binary format (`.bin`)
- Test images (PNG format)

---

## Usage

### 1. Open Tool

```bash
open tools/cnn_v2_test/index.html
```

Or use a local server to avoid CORS:
```bash
python3 -m http.server 8000
# Open http://localhost:8000/tools/cnn_v2_test/
```

### 2. Load Data

1. **Drop `.bin` weights** into left sidebar zone (or click to browse)
2. **Drop PNG image** anywhere in center canvas area
3. CNN runs automatically when both loaded

### 3. Layout

**Left Sidebar:**
- Weights drop zone (click or drag-drop `.bin` files)
- Weights info panel (layer specs, ranges, file size)
- Weights visualization (click Layer 0/1/2 buttons)
  - 1 pixel per weight, all input channels horizontally
  - Output channels (Out 0-3) stacked vertically

**Center Canvas:**
- Main output view (CNN result, original, or diff)
- Keyboard: `SPACE` = original, `D` = diff (×10)

**Right Sidebar:**
- Layer selection buttons (Static 0-3/4-7, Layer 0/1/2)
- 4 small activation views (Ch0/1/2/3) in a row
- Large zoom view below (4× magnification, follows mouse)

**Header Controls:**
- **Blend:** Mix between original (0.0) and CNN output (1.0)
- **Depth:** Uniform depth value for all pixels (0.0–1.0)
- **View:** Current display mode

**Footer:**
- Status: GPU timing (ms), image dimensions, view mode
- Console: Timestamped event log (file loads, errors)

---

## Preparing Test Data

### Export Weights

```bash
# From trained checkpoint
./training/export_cnn_v2_weights.py \
  checkpoints/checkpoint_epoch_100.pth \
  --output-weights tools/cnn_v2_test/test_weights.bin
```

Binary format: 16-byte header + 20 bytes per layer + f16 weights (~3.2 KB for 3-layer model)

### Test Images

Use training images or any PNG:
```bash
# Copy test image
cp training/input/test.png tools/cnn_v2_test/
```

**Note:** Grayscale images automatically converted to RGB.

---

## Validation

### Visual Comparison

Compare browser output with C++ tool:

```bash
# Generate C++ output
./build/cnn_test training/input/test.png /tmp/cpp_output.png

# Load same image in browser tool
# Visually compare outputs
```

### GPU Timing

Expected performance:
- 512×512: ~1-2 ms (integrated GPU)
- 1024×1024: ~3-5 ms
- 1920×1080: ~5-8 ms

Slower than expected? Check:
- WebGPU enabled in browser
- Dedicated GPU selected (if available)
- No background tabs consuming GPU

---

## Troubleshooting

### "WebGPU not supported"

- Update browser to latest version
- Enable WebGPU flag: `chrome://flags/#enable-unsafe-webgpu`
- Try Safari 18+ (native WebGPU on macOS)

### "Invalid .bin file"

- Check magic number: `hexdump -C weights.bin | head`
- Should start with: `43 4e 4e 32` ('CNN2')
- Re-export weights: `./training/export_cnn_v2_weights.py`

### Black output / incorrect colors

- Check blend slider (set to 1.0 for full CNN output)
- Verify training converged (loss < 0.01)
- Compare with C++ tool output

### Shader compilation errors

Open browser console (F12) for detailed errors. Common issues:
- Image too large (>4096×4096 not tested)
- Unsupported texture format (rare on modern GPUs)

---

## Architecture

**Pipeline:**
1. **Static Features Pass** - Generate 8D features (RGBD, UV, sin, bias)
2. **CNN Layer Passes** - Compute N layers with ping-pong textures
3. **Display Pass** - Unpack and render with view mode

**Textures:**
- Input: RGBA8 (original image)
- Depth: R32F (uniform depth)
- Static features: RGBA32Uint (8×f16 packed)
- Layer buffers: RGBA32Uint (ping-pong)

**Data-Driven Execution:**
- Layer count read from binary header
- Per-layer params (kernel size, channels, offsets) from binary
- Single CNN shader dispatched N times

---

## Implemented Features

**✓ Weights Metadata Panel:**
- Layer descriptions (kernel size, channels, weight count)
- Weight statistics (min/max per layer)
- File size and layer count

**✓ Weights Visualization:**
- Per-layer kernel heatmaps (1px/weight)
- All input channels displayed horizontally
- Output channels stacked vertically
- Normalized grayscale display

**✓ Layer Activation Viewer:**
- Static features (8D split into 0-3 and 4-7 views)
- All CNN layer outputs (Layer 0/1/2...)
- 4-channel split view (grayscale per channel)
- Mouse-driven 4× zoom view

## TODO

**Future Enhancements:**
- Weight distribution histograms per layer
- Activation statistics (min/max/mean overlay)
- Side-by-side diff mode (browser vs C++ output)
- Export rendered layers as PNG

---

## Extensions (v2+)

Planned enhancements:

**Variable Feature Count:**
- Binary v2: Add `num_features` to header
- Shader: Dynamic feature array or multiple textures

**Multi-Scale Input (Mip Levels):**
- Uncomment mip bindings in static shader
- No binary format change needed

**8-bit Quantized Weights:**
- Binary version bump (format field already present)
- Add quantization codepath in `get_weight()` function
- 2× size reduction (~1.6 KB)

**Pre-defined Test Images:**
- Dropdown menu with training/input/*.png
- Requires local file server

---

## Size

- HTML structure: ~2 KB
- CSS styling: ~2 KB
- JavaScript logic: ~10 KB (includes zoom + weights viz)
- Static shader: ~1 KB
- CNN shader: ~3 KB
- Display shader: ~1 KB
- Layer viz shader: ~2 KB
- Zoom shader: ~1 KB
- **Total: ~22 KB** (single file, no dependencies)

---

## See Also

- `doc/CNN_V2.md` - Architecture and design
- `doc/HOWTO.md` - Training workflows
- `training/export_cnn_v2_weights.py` - Binary format
- `src/gpu/effects/cnn_v2_effect.cc` - C++ reference implementation