Datasets:

notgoodkeeper
/

cnn-based-drowsiness-detection-data

image stringlengths 45 45	drowsiness_class int64 0 4	drowsiness_score float64 0 1	ear float64 0 1	mar float64 0 1	head_roll float64 0 0.79	head_pitch float64 0 0.88	head_combined float64 0 0.79
images/face_processed_20251007_014055_723.png	3	0.65	0.073324	0.550419	0.086953	0.076872	0.086721
images/face_processed_20251007_013933_471.png	2	0.35	0.082134	0.068425	0.014674	0.003647	0.014069
images/face_processed_20251007_013929_395.png	0	0	0.386738	0.04299	0.011209	0.043227	0.012544
images/face_processed_20251007_013920_881.png	0	0	0.595644	0.014878	0.038518	0.053231	0.038663
images/face_processed_20251007_014106_279.png	2	0.35	0.164624	0.017559	0.105605	0.023187	0.104466
images/face_processed_20251007_014005_544.png	2	0.3875	0.29216	0.596809	0.092954	0.056564	0.092264
images/face_processed_20251007_014030_539.png	0	0	0.670122	0.003688	0.128685	0.206906	0.131444
images/face_processed_20251007_014037_505.png	2	0.35	0.310599	0.014297	0.603419	0.488743	0.603953
images/face_processed_20251007_013950_135.png	0	0.09	0.437554	0.188454	0.044608	0.038285	0.044211
images/face_processed_20251007_013935_507.png	2	0.35	0.11916	0.066839	0.013955	0.040994	0.014784
images/face_processed_20251007_013927_073.png	0	0	0.650575	0.043378	0.044894	0.033253	0.044392
images/face_processed_20251007_013934_649.png	0	0	0.670524	0.054705	0.048565	0.033658	0.048023
images/face_processed_20251007_013952_996.png	0	0	0.437598	0.01453	0.046688	0.038847	0.04627
images/face_processed_20251007_014023_820.png	2	0.35	0.139478	0.016793	0.090484	0.027088	0.089481
images/face_processed_20251007_014009_913.png	2	0.3875	0.278021	0.707816	0.139087	0.029441	0.137746
images/face_processed_20251007_013926_691.png	0	0	0.643503	0.018899	0.037589	0.038138	0.037309
images/face_processed_20251007_013944_272.png	0	0	0.461671	0.112223	0.021186	0.078326	0.023896
images/face_processed_20251007_014005_238.png	2	0.3	0.349955	0.732256	0.088676	0.063588	0.088153
images/face_processed_20251007_013949_920.png	0	0.09	0.426369	0.236809	0.065129	0.017775	0.064248
images/face_processed_20251007_014000_287.png	2	0.3	0.343826	0.813115	0.108736	0.02812	0.107604
images/face_processed_20251007_014048_966.png	0	0	0.503322	0.013753	0.008793	0.069433	0.013497
images/face_processed_20251007_013950_269.png	0	0	0.420006	0.111221	0.050523	0.041081	0.050094
images/face_processed_20251007_014032_668.png	0	0	0.562208	0.018999	0.028833	0.10328	0.032446
images/face_processed_20251007_013922_592.png	2	0.35	0.067587	0.030775	0.069746	0.015573	0.068818
images/face_processed_20251007_014023_014.png	2	0.35	0.105172	0.012841	0.050642	0.052821	0.050482
images/face_processed_20251007_014125_513.png	3	0.65	0.108318	0.504547	0.073732	0.177839	0.077936
images/face_processed_20251007_013909_983.png	0	0	0.513274	0.015266	0.007175	0.007436	0.006706
images/face_processed_20251007_014151_871.png	0	0	0.486216	0.179623	0.06362	0.079707	0.063931
images/face_processed_20251007_013932_814.png	2	0.35	0.020898	0.058535	0.03285	0.049379	0.033038
images/face_processed_20251007_014024_541.png	1	0.175	0.245024	0.020489	0.052547	0.144773	0.056476
images/face_processed_20251007_013959_664.png	2	0.3	0.373897	0.883193	0.065957	0.035879	0.065255
images/face_processed_20251007_014022_084.png	0	0	0.381884	0.014878	0.097406	0.031261	0.096382
images/face_processed_20251007_013955_525.png	2	0.3	0.44516	0.887306	0.107038	0.007809	0.105833
images/face_processed_20251007_013929_617.png	0	0	0.429723	0.054919	0.011257	0.027983	0.011508
images/face_processed_20251007_014155_272.png	2	0.44	0.180186	0.241955	0.106543	0.137721	0.107547
images/face_processed_20251007_014002_005.png	1	0.255	0.392548	0.338669	0.069813	0.054733	0.069379
images/face_processed_20251007_013948_463.png	2	0.3	0.388712	0.656129	0.071828	0.027034	0.070971
images/face_processed_20251007_014153_447.png	1	0.2625	0.226162	0.016694	0.08711	0.049587	0.08638
images/face_processed_20251007_014037_338.png	2	0.35	0.386594	0.012115	0.743967	0.879634	0.751583
images/face_processed_20251007_014154_711.png	2	0.35	0.109087	0.015222	0.05828	0.113108	0.060078
images/face_processed_20251007_013946_736.png	0	0	0.466737	0.118379	0.014821	0.038699	0.015425
images/face_processed_20251007_014124_676.png	4	0.7275	0.16828	0.393937	0.175834	0.086739	0.174715
images/face_processed_20251007_013946_828.png	0	0	0.489799	0.152187	0.013616	0.060283	0.016023
images/face_processed_20251007_014109_500.png	3	0.6125	0.216271	0.147962	0.469846	0.436203	0.471349
images/face_processed_20251007_013925_800.png	0	0	0.657065	0.038666	0.042054	0.0379	0.041689
images/face_processed_20251007_014151_313.png	0	0	0.510649	0.091602	0.113214	0.041372	0.112152
images/face_processed_20251007_014117_734.png	4	0.7375	0.27813	0.450265	0.468032	0.239382	0.466025
images/face_processed_20251007_014010_020.png	2	0.3	0.323243	0.707559	0.124638	0.011402	0.123325
images/face_processed_20251007_014002_048.png	1	0.18	0.388995	0.312248	0.080466	0.04092	0.079687
images/face_processed_20251007_013957_672.png	0	0	0.436972	0.013649	0.048941	0.062454	0.049095
images/face_processed_20251007_014152_050.png	1	0.18	0.48268	0.300916	0.116524	0.050361	0.115524
images/face_processed_20251007_013958_754.png	0	0.09	0.396013	0.239022	0.083941	0.021909	0.082949
images/face_processed_20251007_013922_128.png	1	0.2625	0.234892	0.018584	0.020941	0.039022	0.021181
images/face_processed_20251007_014131_450.png	3	0.53	0.1774	0.300253	0.017458	0.143709	0.028039
images/face_processed_20251007_014202_837.png	3	0.65	0.159742	0.643879	0.031798	0.131276	0.037267
images/face_processed_20251007_014033_418.png	0	0	0.707718	0.027023	0.102971	0.241025	0.108637
images/face_processed_20251007_014000_396.png	2	0.3	0.341224	0.798555	0.091615	0.009151	0.090513
images/face_processed_20251007_014102_108.png	0	0	0.483641	0.065297	0.008146	0.012184	0.0078
images/face_processed_20251007_013918_509.png	0	0	0.54407	0.014851	0.010112	0.027657	0.010437
images/face_processed_20251007_014017_957.png	0	0	0.534586	0.014535	0.004138	0.062341	0.01014
images/face_processed_20251007_014032_712.png	0	0	0.591559	0.020014	0.025187	0.089072	0.028188
images/face_processed_20251007_014156_162.png	1	0.255	0.314846	0.396827	0.065491	0.014933	0.06459
images/face_processed_20251007_014026_150.png	0	0	0.549686	0.022513	0.080092	0.004623	0.079057
images/face_processed_20251007_014007_501.png	0	0	0.363948	0.015971	0.10159	0.019389	0.100459
images/face_processed_20251007_014036_072.png	0	0	0.456123	0.013967	0.116721	0.186572	0.119133
images/face_processed_20251007_013942_560.png	2	0.3	0.455143	0.631289	0.042916	0.006012	0.042127
images/face_processed_20251007_014053_690.png	3	0.65	0.117446	0.443561	0.098654	0.026401	0.097585
images/face_processed_20251007_014031_615.png	0	0	0.490419	0.010205	0.034458	0.129968	0.039397
images/face_processed_20251007_013918_025.png	0	0	0.512581	0.013315	0.015238	0.035543	0.015628
images/face_processed_20251007_014020_369.png	0	0	0.347412	0.01918	0.113439	0.115315	0.113649
images/face_processed_20251007_013930_312.png	0	0	0.537143	0.016036	0.024971	0.007149	0.024314
images/face_processed_20251007_014156_359.png	1	0.265	0.264202	0.240454	0.095344	0.005203	0.094211
images/face_processed_20251007_013954_976.png	2	0.3	0.438332	0.686482	0.047865	0.029929	0.04727
images/face_processed_20251007_013959_471.png	2	0.3	0.372629	0.868313	0.066609	0.070353	0.066586
images/face_processed_20251007_014129_649.png	3	0.5525	0.333246	0.423113	0.322759	0.582423	0.333083
images/face_processed_20251007_013933_288.png	2	0.35	0.028807	0.039891	0.036657	0.032081	0.036251
images/face_processed_20251007_014124_389.png	3	0.57	0.200481	0.38029	0.110418	0.058651	0.109576
images/face_processed_20251007_013928_327.png	2	0.35	0.110092	0.030079	0.018002	0.053688	0.019276
images/face_processed_20251007_014048_702.png	2	0.35	0.152906	0.011924	0.021534	0.054946	0.02257
images/face_processed_20251007_014151_476.png	1	0.255	0.517507	0.376738	0.099179	0.048575	0.098316
images/face_processed_20251007_014035_810.png	0	0	0.426633	0.013684	0.019436	0.141356	0.028993
images/face_processed_20251007_013920_510.png	0	0	0.645898	0.014971	0.055729	0.049671	0.055402
images/face_processed_20251007_013940_985.png	1	0.18	0.455181	0.331437	0.002164	0.043994	0.006741
images/face_processed_20251007_014108_992.png	2	0.4725	0.245977	0.016553	0.422552	0.484614	0.426223
images/face_processed_20251007_014125_264.png	3	0.65	0.120507	0.480032	0.021052	0.20248	0.037681
images/face_processed_20251007_014003_858.png	0	0.09	0.360647	0.20641	0.06627	0.040563	0.065633
images/face_processed_20251007_014027_431.png	0	0	0.53142	0.013864	0.017845	0.047086	0.018708
images/face_processed_20251007_014152_334.png	1	0.255	0.539464	0.340426	0.109023	0.038384	0.107968
images/face_processed_20251007_014125_305.png	3	0.65	0.141564	0.521161	0.021765	0.131974	0.029511
images/face_processed_20251007_013941_660.png	2	0.3	0.43912	0.504984	0.011097	0.009936	0.010616
images/face_processed_20251007_014033_641.png	0	0.1225	0.700328	0.029869	0.199339	0.217519	0.20049
images/face_processed_20251007_013945_906.png	0	0	0.482721	0.013598	0.002867	0.072582	0.011277
images/face_processed_20251007_014125_222.png	3	0.65	0.105225	0.456342	0.043472	0.186565	0.051763
images/face_processed_20251007_014030_346.png	1	0.2275	1	0.005593	0.243148	0.047558	0.241175
images/face_processed_20251007_014053_557.png	1	0.255	0.550688	0.424077	0.026155	0.000406	0.025464
images/face_processed_20251007_014129_155.png	3	0.6925	0.280511	0.425573	0.596754	0.557595	0.598886
images/face_processed_20251007_014049_989.png	0	0	0.493397	0.012519	0.004652	0.032569	0.006394
images/face_processed_20251007_014028_259.png	0	0	0.636658	0.005977	0.087233	0.264486	0.095688
images/face_processed_20251007_014008_317.png	2	0.475	0.242449	0.554116	0.082394	0.028971	0.081469
images/face_processed_20251007_014058_032.png	0	0	0.544828	0.042299	0.006559	0.035139	0.00804

End of preview. Expand in Data Studio

CNN-Based Drowsiness Detection - Dataset

Preprocessed, auto-labeled face-crop images used to train the model in notgoodkeeper/cnn-based-drowsiness-detection.

Code: https://github.com/not-good-keeper/cnn-based-drowsiness-detection

Collection

Frames were captured from a webcam, then run through:

Haar Cascade face detection -> crop + pad + resize to 412x412
MediaPipe Selfie Segmentation -> background replaced with white
CLAHE contrast normalization -> grayscale

MediaPipe FaceMesh (468 landmarks) was used to compute EAR, MAR, and head-pose (roll/pitch) per frame, and a hand-tuned heuristic scorer turned those into a weak drowsiness_class / drowsiness_score label. See src/feature_extraction.py and src/collect_dataset.py in the code repo for the exact logic.

Splits

split	rows
train	1,661
val	356
test	359

Columns

column	description
`image`	relative path to the preprocessed face image (`images/...png`)
`drowsiness_class`	heuristic-labeled class, 0-4 (alert -> extremely drowsy)
`drowsiness_score`	heuristic score, normalized 0-1
`ear`	eye aspect ratio, normalized 0-1
`mar`	mouth aspect ratio, normalized 0-1
`head_roll`, `head_pitch`, `head_combined`	head-pose angles, normalized 0-1

Regression columns are min-max normalized to [0, 1]; see the code repo's src/realtime_inference.py for the original-scale ranges used to denormalize them.

Downloads last month: -