File size: 10,930 Bytes
eb09c29 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 |
#!/usr/bin/env python3
"""
Comprehensive test suite for TimeSformer model implementation.
Tests all components of the video action recognition system.
"""
import logging
import tempfile
import time
from pathlib import Path
from typing import List, Tuple
import numpy as np
import torch
from PIL import Image
# Import the fixed predictor
from predict_fixed import (
read_video_frames,
normalize_frames,
create_timesformer_tensor,
load_model,
predict_actions
)
# Configure logging
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
def create_test_video_frames(num_frames: int = 8, size: Tuple[int, int] = (224, 224)) -> List[Image.Image]:
"""Create synthetic test frames for testing."""
frames = []
for i in range(num_frames):
# Create frames with different colors to simulate motion
hue = int((i / num_frames) * 255)
color = (hue, 255 - hue, 128)
frame = Image.new('RGB', size, color)
frames.append(frame)
return frames
def test_frame_creation():
"""Test synthetic frame creation."""
print("\n🔍 Testing frame creation...")
try:
frames = create_test_video_frames()
assert len(frames) == 8, f"Expected 8 frames, got {len(frames)}"
assert all(frame.size == (224, 224) for frame in frames), "Frame size mismatch"
assert all(frame.mode == 'RGB' for frame in frames), "Frame mode should be RGB"
print("✅ Frame creation test passed")
return True
except Exception as e:
print(f"❌ Frame creation test failed: {e}")
return False
def test_frame_normalization():
"""Test frame normalization function."""
print("\n🔍 Testing frame normalization...")
try:
# Create frames with different sizes
frames = [
Image.new('RGB', (100, 100), 'red'),
Image.new('RGB', (300, 200), 'green'),
Image.new('RGBA', (224, 224), 'blue') # Different mode
]
normalized = normalize_frames(frames, target_size=(224, 224))
assert len(normalized) == 3, "Frame count mismatch"
assert all(frame.size == (224, 224) for frame in normalized), "Normalization size failed"
assert all(frame.mode == 'RGB' for frame in normalized), "Mode conversion failed"
print("✅ Frame normalization test passed")
return True
except Exception as e:
print(f"❌ Frame normalization test failed: {e}")
return False
def test_tensor_creation():
"""Test TimeSformer tensor creation."""
print("\n🔍 Testing TimeSformer tensor creation...")
try:
frames = create_test_video_frames(8)
tensor = create_timesformer_tensor(frames)
# Check tensor properties
expected_shape = (1, 8, 3, 224, 224) # (batch, frames, channels, height, width)
assert tensor.shape == expected_shape, f"Expected shape {expected_shape}, got {tensor.shape}"
assert tensor.dtype == torch.float32, f"Expected float32, got {tensor.dtype}"
assert 0.0 <= tensor.min() <= 1.0, f"Tensor values should be normalized, min: {tensor.min()}"
assert 0.0 <= tensor.max() <= 1.0, f"Tensor values should be normalized, max: {tensor.max()}"
print(f"✅ Tensor creation test passed - Shape: {tensor.shape}")
return True
except Exception as e:
print(f"❌ Tensor creation test failed: {e}")
return False
def test_model_loading():
"""Test model loading functionality."""
print("\n🔍 Testing model loading...")
try:
processor, model, device = load_model()
# Check model properties
assert processor is not None, "Processor should not be None"
assert model is not None, "Model should not be None"
assert hasattr(model, 'config'), "Model should have config"
assert hasattr(model.config, 'id2label'), "Model should have label mapping"
# Check if model is in eval mode
assert not model.training, "Model should be in eval mode"
# Check device
model_device = next(model.parameters()).device
print(f"Model loaded on device: {model_device}")
print("✅ Model loading test passed")
return True
except Exception as e:
print(f"❌ Model loading test failed: {e}")
return False
def test_end_to_end_prediction():
"""Test complete prediction pipeline with synthetic video."""
print("\n🔍 Testing end-to-end prediction...")
try:
# Create a temporary video file (we'll simulate this with frames)
frames = create_test_video_frames(8)
# Create temporary directory and mock video processing
with tempfile.TemporaryDirectory() as temp_dir:
# We'll test the tensor creation and model inference directly
# since creating an actual video file is complex
# Test tensor creation
tensor = create_timesformer_tensor(frames)
# Load model
processor, model, device = load_model()
# Move tensor to device
tensor = tensor.to(device)
# Run inference
with torch.no_grad():
outputs = model(pixel_values=tensor)
logits = outputs.logits
# Check output properties
assert logits.shape[0] == 1, "Batch size should be 1"
assert logits.shape[1] == 400, "Should have 400 classes (Kinetics-400)"
# Get top predictions
probabilities = torch.softmax(logits, dim=-1)[0]
top_probs, top_indices = torch.topk(probabilities, k=5)
# Convert to results
results = []
for prob, idx in zip(top_probs.cpu(), top_indices.cpu()):
label = model.config.id2label[idx.item()]
confidence = float(prob.item())
results.append((label, confidence))
# Validate results
assert len(results) == 5, "Should return 5 predictions"
assert all(isinstance(label, str) for label, _ in results), "Labels should be strings"
assert all(0.0 <= confidence <= 1.0 for _, confidence in results), "Confidence should be between 0 and 1"
assert all(results[i][1] >= results[i+1][1] for i in range(len(results)-1)), "Results should be sorted by confidence"
print("✅ End-to-end prediction test passed")
print(f"Top prediction: {results[0][0]} ({results[0][1]:.4f})")
return True
except Exception as e:
print(f"❌ End-to-end prediction test failed: {e}")
return False
def test_error_handling():
"""Test error handling scenarios."""
print("\n🔍 Testing error handling...")
tests_passed = 0
total_tests = 3
# Test 1: Invalid number of frames
try:
frames = create_test_video_frames(5) # Wrong number
create_timesformer_tensor(frames)
print("❌ Should have failed with wrong frame count")
except ValueError:
print("✅ Correctly handled wrong frame count")
tests_passed += 1
except Exception as e:
print(f"❌ Unexpected error for wrong frame count: {e}")
# Test 2: Empty frame list
try:
normalize_frames([])
print("❌ Should have failed with empty frames")
except (RuntimeError, ValueError):
print("✅ Correctly handled empty frame list")
tests_passed += 1
except Exception as e:
print(f"❌ Unexpected error for empty frames: {e}")
# Test 3: Invalid frame type
try:
frames = [None] * 8
create_timesformer_tensor(frames)
print("❌ Should have failed with invalid frame type")
except (AttributeError, TypeError):
print("✅ Correctly handled invalid frame type")
tests_passed += 1
except Exception as e:
print(f"❌ Unexpected error for invalid frames: {e}")
success_rate = tests_passed / total_tests
print(f"Error handling tests: {tests_passed}/{total_tests} passed ({success_rate:.1%})")
return success_rate >= 0.8
def benchmark_performance():
"""Benchmark the performance of key operations."""
print("\n⏱️ Benchmarking performance...")
# Benchmark tensor creation
frames = create_test_video_frames(8)
start_time = time.time()
for _ in range(10):
tensor = create_timesformer_tensor(frames)
tensor_time = (time.time() - start_time) / 10
print(f"Average tensor creation time: {tensor_time:.4f} seconds")
# Benchmark model inference
try:
processor, model, device = load_model()
tensor = create_timesformer_tensor(frames).to(device)
# Warm up
with torch.no_grad():
model(pixel_values=tensor)
# Benchmark
start_time = time.time()
for _ in range(5):
with torch.no_grad():
outputs = model(pixel_values=tensor)
inference_time = (time.time() - start_time) / 5
print(f"Average model inference time: {inference_time:.4f} seconds")
print(f"Device used: {device}")
if tensor_time < 0.1 and inference_time < 2.0:
print("✅ Performance benchmarks look good")
return True
else:
print("⚠️ Performance might be slower than expected")
return True # Don't fail on slow performance
except Exception as e:
print(f"❌ Benchmark failed: {e}")
return False
def run_all_tests():
"""Run all tests and provide summary."""
print("🚀 Starting TimeSformer Model Test Suite")
print("=" * 60)
tests = [
("Frame Creation", test_frame_creation),
("Frame Normalization", test_frame_normalization),
("Tensor Creation", test_tensor_creation),
("Model Loading", test_model_loading),
("End-to-End Prediction", test_end_to_end_prediction),
("Error Handling", test_error_handling),
("Performance Benchmark", benchmark_performance),
]
passed = 0
total = len(tests)
for test_name, test_func in tests:
try:
if test_func():
passed += 1
else:
print(f"💥 {test_name} failed")
except Exception as e:
print(f"💥 {test_name} crashed: {e}")
print("\n" + "=" * 60)
print(f"📊 TEST SUMMARY: {passed}/{total} tests passed ({passed/total:.1%})")
if passed == total:
print("🎉 ALL TESTS PASSED! Your TimeSformer implementation is working correctly.")
elif passed >= total * 0.8:
print("✅ Most tests passed. Minor issues may exist but the core functionality works.")
else:
print("❌ Several tests failed. Please review the implementation.")
return passed == total
if __name__ == "__main__":
success = run_all_tests()
exit(0 if success else 1)
|