Hugging Face's logo

CaptainRapid
/
acoustic-alert-5class

Audio Classification
Keras
tensorflow
embedded-ai
raspberry-pi
acoustic-event-detection
Model card Files
xet
 Community
CaptainRapid commited on
Commit
23729a3
·
verified ·
1 Parent(s): 9f09fb9

Upload 5-class acoustic alert baseline

Browse files
Files changed (11) hide show
  1. .gitattributes +3 -0
  2. README.md +107 -0
  3. best_model.keras +3 -0
  4. classification_report.txt +11 -0
  5. confidence_threshold_sweep.json +444 -0
  6. confusion_matrix.png +3 -0
  7. final_model.keras +3 -0
  8. label_map.json +36 -0
  9. metrics.json +91 -0
  10. normalization.json +17 -0
  11. training_summary.json +20 -0
.gitattributes CHANGED
@@ -33,3 +33,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ best_model.keras filter=lfs diff=lfs merge=lfs -text
37
+ confusion_matrix.png filter=lfs diff=lfs merge=lfs -text
38
+ final_model.keras filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ library_name: keras
4
+ tags:
5
+ - audio-classification
6
+ - tensorflow
7
+ - keras
8
+ - embedded-ai
9
+ - raspberry-pi
10
+ - acoustic-event-detection
11
+ datasets:
12
+ - karolpiczak/ESC-50
13
+ - danavery/urbansound8K
14
+ - FSD50K
15
+ metrics:
16
+ - accuracy
17
+ - f1
18
+ ---
19
+
20
+ # On-device Acoustic Alert System - 5-Class Baseline
21
+
22
+ This repository contains the current lightweight CNN baseline for a Raspberry Pi on-device acoustic alert system.
23
+
24
+ ## Classes
25
+
26
+ ```text
27
+ background
28
+ dog_bark
29
+ alarm_siren
30
+ door_event
31
+ scream_shout
32
+ ```
33
+
34
+ `baby_cry` and `glass_break` were removed from this baseline because the available data was insufficient or unstable for a reliable first deployable model.
35
+
36
+ ## Model
37
+
38
+ - Framework: TensorFlow/Keras
39
+ - Architecture: Lightweight CNN
40
+ - Input: normalized Log-Mel Spectrogram, shape `(64, 64, 1)`
41
+ - Parameters: about 28k
42
+ - Intended deployment path: TensorFlow Lite / INT8 quantization in the next project stage
43
+
44
+ ## Feature Extraction
45
+
46
+ Training and inference must use the same parameters:
47
+
48
+ ```text
49
+ sample_rate = 16000
50
+ clip_seconds = 2
51
+ n_mels = 64
52
+ n_fft = 1024
53
+ hop_length = 512
54
+ ```
55
+
56
+ Normalization parameters are stored in `normalization.json`.
57
+
58
+ ## Evaluation
59
+
60
+ Test results for `best_model.keras`:
61
+
62
+ | metric | value |
63
+ | --- | ---: |
64
+ | accuracy | 0.6954 |
65
+ | macro F1 | 0.5402 |
66
+ | mean inference latency | 43.88 ms |
67
+
68
+ Per-class F1:
69
+
70
+ | class | F1 |
71
+ | --- | ---: |
72
+ | background | 0.7962 |
73
+ | dog_bark | 0.4806 |
74
+ | alarm_siren | 0.5023 |
75
+ | door_event | 0.4579 |
76
+ | scream_shout | 0.4638 |
77
+
78
+ Confidence-threshold behavior:
79
+
80
+ | threshold | accuracy | macro F1 | background false-alert rate | target detection rate |
81
+ | --- | ---: | ---: | ---: | ---: |
82
+ | 0.65 | 0.7803 | 0.6105 | 0.1378 | 0.5801 |
83
+ | 0.75 | 0.7970 | 0.6191 | 0.0976 | 0.5096 |
84
+ | 0.80 | 0.8035 | 0.6184 | 0.0788 | 0.4720 |
85
+
86
+ For long-running alerting, use thresholding plus alert smoothing and cooldown.
87
+
88
+ ## Files
89
+
90
+ - `best_model.keras`: best validation checkpoint
91
+ - `final_model.keras`: final training checkpoint
92
+ - `label_map.json`: label order used by model outputs
93
+ - `normalization.json`: feature normalization and extraction parameters
94
+ - `metrics.json`: test metrics
95
+ - `classification_report.txt`: per-class report
96
+ - `confidence_threshold_sweep.json`: threshold trade-off table
97
+ - `confusion_matrix.png`: confusion matrix plot
98
+ - `training_summary.json`: training run summary
99
+
100
+ ## Limitations
101
+
102
+ This is a course-project baseline, not a production safety system.
103
+
104
+ - It does not save or upload raw audio.
105
+ - It does not perform speech recognition.
106
+ - It detects sound event categories only.
107
+ - Performance on Raspberry Pi microphone audio still needs device-side validation.
best_model.keras ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cfb7b535336f39258a7ff6d3d9436a1a390fdfd0fcbf9a84dc0054064f763ff8
3
+ size 407139
classification_report.txt ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ precision recall f1-score support
2
+
3
+ background 0.90 0.71 0.80 3831
4
+ dog_bark 0.36 0.71 0.48 226
5
+ alarm_siren 0.44 0.59 0.50 635
6
+ door_event 0.32 0.78 0.46 111
7
+ scream_shout 0.37 0.62 0.46 276
8
+
9
+ accuracy 0.70 5079
10
+ macro avg 0.48 0.68 0.54 5079
11
+ weighted avg 0.78 0.70 0.72 5079
confidence_threshold_sweep.json ADDED
@@ -0,0 +1,444 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "split": "test",
3
+ "labels": [
4
+ "background",
5
+ "dog_bark",
6
+ "alarm_siren",
7
+ "door_event",
8
+ "scream_shout"
9
+ ],
10
+ "background_id": 0,
11
+ "threshold_policy": "If top prediction is a non-background class and confidence is below threshold, output background.",
12
+ "thresholds": {
13
+ "0.0": {
14
+ "accuracy": 0.6954124827721992,
15
+ "macro_precision": 0.4785615211327133,
16
+ "macro_recall": 0.683787415881107,
17
+ "macro_f1": 0.5401677278598453,
18
+ "background_false_alerts": 1093,
19
+ "background_false_alert_rate": 0.285304098146698,
20
+ "target_hits": 939,
21
+ "target_misses_to_background": 309,
22
+ "target_detection_rate": 0.7524038461538461,
23
+ "per_class": {
24
+ "background": {
25
+ "precision": 0.8985887758450936,
26
+ "recall": 0.714695901853302,
27
+ "f1": 0.7961616749054958,
28
+ "support": 3831
29
+ },
30
+ "dog_bark": {
31
+ "precision": 0.36261261261261263,
32
+ "recall": 0.7123893805309734,
33
+ "f1": 0.48059701492537316,
34
+ "support": 226
35
+ },
36
+ "alarm_siren": {
37
+ "precision": 0.4361948955916473,
38
+ "recall": 0.5921259842519685,
39
+ "f1": 0.5023380093520374,
40
+ "support": 635
41
+ },
42
+ "door_event": {
43
+ "precision": 0.32342007434944237,
44
+ "recall": 0.7837837837837838,
45
+ "f1": 0.45789473684210524,
46
+ "support": 111
47
+ },
48
+ "scream_shout": {
49
+ "precision": 0.37199124726477023,
50
+ "recall": 0.6159420289855072,
51
+ "f1": 0.4638472032742155,
52
+ "support": 276
53
+ }
54
+ }
55
+ },
56
+ "0.4": {
57
+ "accuracy": 0.7050600511911793,
58
+ "macro_precision": 0.4881460770540735,
59
+ "macro_recall": 0.6836289748301112,
60
+ "macro_f1": 0.5495130654668229,
61
+ "background_false_alerts": 1038,
62
+ "background_false_alert_rate": 0.27094753328112764,
63
+ "target_hits": 917,
64
+ "target_misses_to_background": 331,
65
+ "target_detection_rate": 0.7347756410256411,
66
+ "per_class": {
67
+ "background": {
68
+ "precision": 0.8940460947503202,
69
+ "recall": 0.7290524667188724,
70
+ "f1": 0.8031631919482387,
71
+ "support": 3831
72
+ },
73
+ "dog_bark": {
74
+ "precision": 0.37857142857142856,
75
+ "recall": 0.7035398230088495,
76
+ "f1": 0.49226006191950467,
77
+ "support": 226
78
+ },
79
+ "alarm_siren": {
80
+ "precision": 0.4449760765550239,
81
+ "recall": 0.5858267716535434,
82
+ "f1": 0.5057783820530252,
83
+ "support": 635
84
+ },
85
+ "door_event": {
86
+ "precision": 0.33852140077821014,
87
+ "recall": 0.7837837837837838,
88
+ "f1": 0.47282608695652173,
89
+ "support": 111
90
+ },
91
+ "scream_shout": {
92
+ "precision": 0.38461538461538464,
93
+ "recall": 0.6159420289855072,
94
+ "f1": 0.4735376044568245,
95
+ "support": 276
96
+ }
97
+ }
98
+ },
99
+ "0.5": {
100
+ "accuracy": 0.7395156526875369,
101
+ "macro_precision": 0.5322276220267105,
102
+ "macro_recall": 0.670634249081814,
103
+ "macro_f1": 0.584966696328538,
104
+ "background_false_alerts": 825,
105
+ "background_false_alert_rate": 0.2153484729835552,
106
+ "target_hits": 849,
107
+ "target_misses_to_background": 399,
108
+ "target_detection_rate": 0.6802884615384616,
109
+ "per_class": {
110
+ "background": {
111
+ "precision": 0.8828193832599119,
112
+ "recall": 0.7846515270164448,
113
+ "f1": 0.8308457711442786,
114
+ "support": 3831
115
+ },
116
+ "dog_bark": {
117
+ "precision": 0.43641618497109824,
118
+ "recall": 0.668141592920354,
119
+ "f1": 0.527972027972028,
120
+ "support": 226
121
+ },
122
+ "alarm_siren": {
123
+ "precision": 0.4666666666666667,
124
+ "recall": 0.5511811023622047,
125
+ "f1": 0.5054151624548736,
126
+ "support": 635
127
+ },
128
+ "door_event": {
129
+ "precision": 0.45604395604395603,
130
+ "recall": 0.7477477477477478,
131
+ "f1": 0.5665529010238908,
132
+ "support": 111
133
+ },
134
+ "scream_shout": {
135
+ "precision": 0.41919191919191917,
136
+ "recall": 0.6014492753623188,
137
+ "f1": 0.49404761904761907,
138
+ "support": 276
139
+ }
140
+ }
141
+ },
142
+ "0.6": {
143
+ "accuracy": 0.7676708013388462,
144
+ "macro_precision": 0.5719646609445405,
145
+ "macro_recall": 0.6502101795683431,
146
+ "macro_f1": 0.6046756023804498,
147
+ "background_false_alerts": 623,
148
+ "background_false_alert_rate": 0.16262072565909685,
149
+ "target_hits": 763,
150
+ "target_misses_to_background": 485,
151
+ "target_detection_rate": 0.6113782051282052,
152
+ "per_class": {
153
+ "background": {
154
+ "precision": 0.8686704576225291,
155
+ "recall": 0.8373792743409032,
156
+ "f1": 0.8527379053694844,
157
+ "support": 3831
158
+ },
159
+ "dog_bark": {
160
+ "precision": 0.5272727272727272,
161
+ "recall": 0.6415929203539823,
162
+ "f1": 0.5788423153692615,
163
+ "support": 226
164
+ },
165
+ "alarm_siren": {
166
+ "precision": 0.5140495867768595,
167
+ "recall": 0.48976377952755906,
168
+ "f1": 0.5016129032258064,
169
+ "support": 635
170
+ },
171
+ "door_event": {
172
+ "precision": 0.5031446540880503,
173
+ "recall": 0.7207207207207207,
174
+ "f1": 0.5925925925925926,
175
+ "support": 111
176
+ },
177
+ "scream_shout": {
178
+ "precision": 0.44668587896253603,
179
+ "recall": 0.5615942028985508,
180
+ "f1": 0.49759229534510435,
181
+ "support": 276
182
+ }
183
+ }
184
+ },
185
+ "0.65": {
186
+ "accuracy": 0.7802717070289427,
187
+ "macro_precision": 0.5922215887840118,
188
+ "macro_recall": 0.6373473645630556,
189
+ "macro_f1": 0.6105274551560768,
190
+ "background_false_alerts": 528,
191
+ "background_false_alert_rate": 0.13782302270947533,
192
+ "target_hits": 724,
193
+ "target_misses_to_background": 524,
194
+ "target_detection_rate": 0.5801282051282052,
195
+ "per_class": {
196
+ "background": {
197
+ "precision": 0.8630781290828325,
198
+ "recall": 0.8621769772905247,
199
+ "f1": 0.8626273178375555,
200
+ "support": 3831
201
+ },
202
+ "dog_bark": {
203
+ "precision": 0.5697211155378487,
204
+ "recall": 0.6327433628318584,
205
+ "f1": 0.59958071278826,
206
+ "support": 226
207
+ },
208
+ "alarm_siren": {
209
+ "precision": 0.5449438202247191,
210
+ "recall": 0.4582677165354331,
211
+ "f1": 0.49786142001710865,
212
+ "support": 635
213
+ },
214
+ "door_event": {
215
+ "precision": 0.5133333333333333,
216
+ "recall": 0.6936936936936937,
217
+ "f1": 0.5900383141762452,
218
+ "support": 111
219
+ },
220
+ "scream_shout": {
221
+ "precision": 0.47003154574132494,
222
+ "recall": 0.5398550724637681,
223
+ "f1": 0.5025295109612141,
224
+ "support": 276
225
+ }
226
+ }
227
+ },
228
+ "0.7": {
229
+ "accuracy": 0.7912974995077772,
230
+ "macro_precision": 0.6194682599995388,
231
+ "macro_recall": 0.6247916951729758,
232
+ "macro_f1": 0.6183632029446662,
233
+ "background_false_alerts": 438,
234
+ "background_false_alert_rate": 0.11433046202036022,
235
+ "target_hits": 679,
236
+ "target_misses_to_background": 569,
237
+ "target_detection_rate": 0.5440705128205128,
238
+ "per_class": {
239
+ "background": {
240
+ "precision": 0.8563856638061585,
241
+ "recall": 0.8856695379796398,
242
+ "f1": 0.870781470550494,
243
+ "support": 3831
244
+ },
245
+ "dog_bark": {
246
+ "precision": 0.6008771929824561,
247
+ "recall": 0.6061946902654868,
248
+ "f1": 0.6035242290748899,
249
+ "support": 226
250
+ },
251
+ "alarm_siren": {
252
+ "precision": 0.5788336933045356,
253
+ "recall": 0.4220472440944882,
254
+ "f1": 0.48816029143898,
255
+ "support": 635
256
+ },
257
+ "door_event": {
258
+ "precision": 0.562962962962963,
259
+ "recall": 0.6846846846846847,
260
+ "f1": 0.6178861788617886,
261
+ "support": 111
262
+ },
263
+ "scream_shout": {
264
+ "precision": 0.49828178694158076,
265
+ "recall": 0.5253623188405797,
266
+ "f1": 0.5114638447971781,
267
+ "support": 276
268
+ }
269
+ }
270
+ },
271
+ "0.75": {
272
+ "accuracy": 0.7970072848986021,
273
+ "macro_precision": 0.6391869562730029,
274
+ "macro_recall": 0.610572168240029,
275
+ "macro_f1": 0.6190753438948531,
276
+ "background_false_alerts": 374,
277
+ "background_false_alert_rate": 0.09762464108587836,
278
+ "target_hits": 636,
279
+ "target_misses_to_background": 612,
280
+ "target_detection_rate": 0.5096153846153846,
281
+ "per_class": {
282
+ "background": {
283
+ "precision": 0.8495944949619071,
284
+ "recall": 0.9023753589141217,
285
+ "f1": 0.8751898734177215,
286
+ "support": 3831
287
+ },
288
+ "dog_bark": {
289
+ "precision": 0.6470588235294118,
290
+ "recall": 0.584070796460177,
291
+ "f1": 0.6139534883720931,
292
+ "support": 226
293
+ },
294
+ "alarm_siren": {
295
+ "precision": 0.5995145631067961,
296
+ "recall": 0.3889763779527559,
297
+ "f1": 0.47182425978987586,
298
+ "support": 635
299
+ },
300
+ "door_event": {
301
+ "precision": 0.5846153846153846,
302
+ "recall": 0.6846846846846847,
303
+ "f1": 0.6307053941908713,
304
+ "support": 111
305
+ },
306
+ "scream_shout": {
307
+ "precision": 0.5151515151515151,
308
+ "recall": 0.4927536231884058,
309
+ "f1": 0.5037037037037037,
310
+ "support": 276
311
+ }
312
+ }
313
+ },
314
+ "0.8": {
315
+ "accuracy": 0.8035046268950581,
316
+ "macro_precision": 0.6632338597788647,
317
+ "macro_recall": 0.595733588505752,
318
+ "macro_f1": 0.6183674611765237,
319
+ "background_false_alerts": 302,
320
+ "background_false_alert_rate": 0.07883059253458627,
321
+ "target_hits": 589,
322
+ "target_misses_to_background": 659,
323
+ "target_detection_rate": 0.4719551282051282,
324
+ "per_class": {
325
+ "background": {
326
+ "precision": 0.8426456542502387,
327
+ "recall": 0.9211694074654138,
328
+ "f1": 0.8801596209003616,
329
+ "support": 3831
330
+ },
331
+ "dog_bark": {
332
+ "precision": 0.6735751295336787,
333
+ "recall": 0.5752212389380531,
334
+ "f1": 0.6205250596658711,
335
+ "support": 226
336
+ },
337
+ "alarm_siren": {
338
+ "precision": 0.6470588235294118,
339
+ "recall": 0.3464566929133858,
340
+ "f1": 0.4512820512820513,
341
+ "support": 635
342
+ },
343
+ "door_event": {
344
+ "precision": 0.6147540983606558,
345
+ "recall": 0.6756756756756757,
346
+ "f1": 0.6437768240343348,
347
+ "support": 111
348
+ },
349
+ "scream_shout": {
350
+ "precision": 0.538135593220339,
351
+ "recall": 0.4601449275362319,
352
+ "f1": 0.49609375,
353
+ "support": 276
354
+ }
355
+ }
356
+ },
357
+ "0.85": {
358
+ "accuracy": 0.8064579641661744,
359
+ "macro_precision": 0.690624826615986,
360
+ "macro_recall": 0.5750611724190517,
361
+ "macro_f1": 0.6136025512101229,
362
+ "background_false_alerts": 239,
363
+ "background_false_alert_rate": 0.06238580005220569,
364
+ "target_hits": 535,
365
+ "target_misses_to_background": 713,
366
+ "target_detection_rate": 0.4286858974358974,
367
+ "per_class": {
368
+ "background": {
369
+ "precision": 0.8343786295005807,
370
+ "recall": 0.9376141999477943,
371
+ "f1": 0.8829891838741396,
372
+ "support": 3831
373
+ },
374
+ "dog_bark": {
375
+ "precision": 0.7102272727272727,
376
+ "recall": 0.5530973451327433,
377
+ "f1": 0.6218905472636815,
378
+ "support": 226
379
+ },
380
+ "alarm_siren": {
381
+ "precision": 0.6798561151079137,
382
+ "recall": 0.29763779527559053,
383
+ "f1": 0.4140197152245345,
384
+ "support": 635
385
+ },
386
+ "door_event": {
387
+ "precision": 0.6788990825688074,
388
+ "recall": 0.6666666666666666,
389
+ "f1": 0.6727272727272727,
390
+ "support": 111
391
+ },
392
+ "scream_shout": {
393
+ "precision": 0.5497630331753555,
394
+ "recall": 0.42028985507246375,
395
+ "f1": 0.47638603696098564,
396
+ "support": 276
397
+ }
398
+ }
399
+ },
400
+ "0.9": {
401
+ "accuracy": 0.8137428627682615,
402
+ "macro_precision": 0.7460151376744764,
403
+ "macro_recall": 0.5543167014058634,
404
+ "macro_f1": 0.61228302931239,
405
+ "background_false_alerts": 154,
406
+ "background_false_alert_rate": 0.04019838162359697,
407
+ "target_hits": 475,
408
+ "target_misses_to_background": 773,
409
+ "target_detection_rate": 0.38060897435897434,
410
+ "per_class": {
411
+ "background": {
412
+ "precision": 0.8262921348314607,
413
+ "recall": 0.959801618376403,
414
+ "f1": 0.8880569979471078,
415
+ "support": 3831
416
+ },
417
+ "dog_bark": {
418
+ "precision": 0.7908496732026143,
419
+ "recall": 0.5353982300884956,
420
+ "f1": 0.6385224274406333,
421
+ "support": 226
422
+ },
423
+ "alarm_siren": {
424
+ "precision": 0.7699530516431925,
425
+ "recall": 0.25826771653543307,
426
+ "f1": 0.3867924528301887,
427
+ "support": 635
428
+ },
429
+ "door_event": {
430
+ "precision": 0.7551020408163265,
431
+ "recall": 0.6666666666666666,
432
+ "f1": 0.7081339712918661,
433
+ "support": 111
434
+ },
435
+ "scream_shout": {
436
+ "precision": 0.5878787878787879,
437
+ "recall": 0.35144927536231885,
438
+ "f1": 0.4399092970521542,
439
+ "support": 276
440
+ }
441
+ }
442
+ }
443
+ }
444
+ }
confusion_matrix.png ADDED

Git LFS Details

  • SHA256: e9ed30723323ac2ba14357184060e7ea709dc9b4321cee62aaec19efd25617e1
  • Pointer size: 131 Bytes
  • Size of remote file: 102 kB
final_model.keras ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eaf2581dd8a008c83063882b48dabe7592f3bc2355f46fa9c9fbdc3799ec85e8
3
+ size 407139
label_map.json ADDED
@@ -0,0 +1,36 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "labels": [
3
+ "background",
4
+ "dog_bark",
5
+ "alarm_siren",
6
+ "door_event",
7
+ "scream_shout"
8
+ ],
9
+ "label_to_id": {
10
+ "background": 0,
11
+ "dog_bark": 1,
12
+ "alarm_siren": 2,
13
+ "door_event": 3,
14
+ "scream_shout": 4
15
+ },
16
+ "id_to_label": {
17
+ "0": "background",
18
+ "1": "dog_bark",
19
+ "2": "alarm_siren",
20
+ "3": "door_event",
21
+ "4": "scream_shout"
22
+ },
23
+ "missing_labels": [
24
+ "baby_cry",
25
+ "glass_break"
26
+ ],
27
+ "canonical_labels": [
28
+ "background",
29
+ "baby_cry",
30
+ "dog_bark",
31
+ "alarm_siren",
32
+ "door_event",
33
+ "glass_break",
34
+ "scream_shout"
35
+ ]
36
+ }
metrics.json ADDED
@@ -0,0 +1,91 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "split": "test",
3
+ "accuracy": 0.6954124827721992,
4
+ "macro_precision": 0.4785615211327133,
5
+ "macro_recall": 0.683787415881107,
6
+ "macro_f1": 0.5401677278598453,
7
+ "per_class": {
8
+ "background": {
9
+ "precision": 0.8985887758450936,
10
+ "recall": 0.714695901853302,
11
+ "f1": 0.7961616749054958,
12
+ "support": 3831
13
+ },
14
+ "dog_bark": {
15
+ "precision": 0.36261261261261263,
16
+ "recall": 0.7123893805309734,
17
+ "f1": 0.48059701492537316,
18
+ "support": 226
19
+ },
20
+ "alarm_siren": {
21
+ "precision": 0.4361948955916473,
22
+ "recall": 0.5921259842519685,
23
+ "f1": 0.5023380093520374,
24
+ "support": 635
25
+ },
26
+ "door_event": {
27
+ "precision": 0.32342007434944237,
28
+ "recall": 0.7837837837837838,
29
+ "f1": 0.45789473684210524,
30
+ "support": 111
31
+ },
32
+ "scream_shout": {
33
+ "precision": 0.37199124726477023,
34
+ "recall": 0.6159420289855072,
35
+ "f1": 0.4638472032742155,
36
+ "support": 276
37
+ }
38
+ },
39
+ "confusion_matrix": [
40
+ [
41
+ 2738,
42
+ 229,
43
+ 460,
44
+ 159,
45
+ 245
46
+ ],
47
+ [
48
+ 34,
49
+ 161,
50
+ 11,
51
+ 6,
52
+ 14
53
+ ],
54
+ [
55
+ 193,
56
+ 24,
57
+ 376,
58
+ 14,
59
+ 28
60
+ ],
61
+ [
62
+ 16,
63
+ 7,
64
+ 1,
65
+ 87,
66
+ 0
67
+ ],
68
+ [
69
+ 66,
70
+ 23,
71
+ 14,
72
+ 3,
73
+ 170
74
+ ]
75
+ ],
76
+ "inference_latency": {
77
+ "mean_ms": 43.87880890397355,
78
+ "median_ms": 43.45465847291052,
79
+ "p95_ms": 46.91382271703333,
80
+ "min_ms": 40.953397983685136,
81
+ "max_ms": 48.14268997870386
82
+ },
83
+ "num_samples": 5079,
84
+ "labels": [
85
+ "background",
86
+ "dog_bark",
87
+ "alarm_siren",
88
+ "door_event",
89
+ "scream_shout"
90
+ ]
91
+ }
normalization.json ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "mean": -41.9937364935039,
3
+ "std": 17.41233292211895,
4
+ "feature_shape": [
5
+ 64,
6
+ 64
7
+ ],
8
+ "sample_rate": 16000,
9
+ "clip_seconds": 2.0,
10
+ "n_mels": 64,
11
+ "n_fft": 1024,
12
+ "hop_length": 512,
13
+ "fmin": 50.0,
14
+ "fmax": 8000.0,
15
+ "power_to_db_ref": "np.max",
16
+ "feature_backend": "soundfile_scipy"
17
+ }
training_summary.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "labels": [
3
+ "background",
4
+ "dog_bark",
5
+ "alarm_siren",
6
+ "door_event",
7
+ "scream_shout"
8
+ ],
9
+ "num_classes": 5,
10
+ "train_samples": 22105,
11
+ "val_samples": 2652,
12
+ "best_val_loss": 0.7007942795753479,
13
+ "best_val_accuracy": 0.7258672714233398,
14
+ "input_shape": [
15
+ 64,
16
+ 64,
17
+ 1
18
+ ],
19
+ "epochs_ran": 60
20
+ }