Hugging Face's logo

Datasets:
SharoonArshad
/
asl_84classes_408d_final

Tasks:
Video Classification
Modalities:
Tabular
Formats:
csv
Size:
1K - 10K
Tags:
asl
sign-language
mediapipe
holistic
keypoints
isolated-sign-recognition
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
Dataset Viewer
Auto-converted to Parquet Duplicate
Search is not available for this dataset
sample_index
int64
0
6.77k
label
int64
0
83
valid_frame_ratio
float64
1
1
zero_frame_ratio
float64
0
0
is_bad
bool
1 class
0
67
1
0
false
1
9
1
0
false
2
18
1
0
false
3
35
1
0
false
4
37
1
0
false
5
41
1
0
false
6
75
1
0
false
7
71
1
0
false
8
20
1
0
false
9
60
1
0
false
10
36
1
0
false
11
24
1
0
false
12
55
1
0
false
13
78
1
0
false
14
67
1
0
false
15
44
1
0
false
16
30
1
0
false
17
29
1
0
false
18
16
1
0
false
19
18
1
0
false
20
31
1
0
false
21
47
1
0
false
22
61
1
0
false
23
75
1
0
false
24
33
1
0
false
25
22
1
0
false
26
54
1
0
false
27
13
1
0
false
28
61
1
0
false
29
78
1
0
false
30
46
1
0
false
31
63
1
0
false
32
7
1
0
false
33
12
1
0
false
34
43
1
0
false
35
66
1
0
false
36
52
1
0
false
37
1
1
0
false
38
83
1
0
false
39
59
1
0
false
40
23
1
0
false
41
34
1
0
false
42
57
1
0
false
43
20
1
0
false
44
52
1
0
false
45
28
1
0
false
46
46
1
0
false
47
35
1
0
false
48
56
1
0
false
49
80
1
0
false
50
30
1
0
false
51
53
1
0
false
52
57
1
0
false
53
78
1
0
false
54
44
1
0
false
55
73
1
0
false
56
12
1
0
false
57
45
1
0
false
58
64
1
0
false
59
20
1
0
false
60
74
1
0
false
61
22
1
0
false
62
54
1
0
false
63
14
1
0
false
64
31
1
0
false
65
60
1
0
false
66
4
1
0
false
67
3
1
0
false
68
55
1
0
false
69
51
1
0
false
70
59
1
0
false
71
68
1
0
false
72
81
1
0
false
73
7
1
0
false
74
27
1
0
false
75
78
1
0
false
76
15
1
0
false
77
27
1
0
false
78
43
1
0
false
79
50
1
0
false
80
29
1
0
false
81
38
1
0
false
82
8
1
0
false
83
81
1
0
false
84
55
1
0
false
85
49
1
0
false
86
52
1
0
false
87
28
1
0
false
88
37
1
0
false
89
34
1
0
false
90
15
1
0
false
91
23
1
0
false
92
66
1
0
false
93
73
1
0
false
94
37
1
0
false
95
64
1
0
false
96
3
1
0
false
97
34
1
0
false
98
16
1
0
false
99
74
1
0
false
End of preview. Expand in Data Studio

YAML Metadata Warning:The task_categories "time-series-classification" is not in the official list: text-classification, token-classification, table-question-answering, question-answering, zero-shot-classification, translation, summarization, feature-extraction, text-generation, fill-mask, sentence-similarity, text-to-speech, text-to-audio, automatic-speech-recognition, audio-to-audio, audio-classification, audio-text-to-text, voice-activity-detection, depth-estimation, image-classification, object-detection, image-segmentation, text-to-image, image-to-text, image-to-image, image-to-video, unconditional-image-generation, video-classification, reinforcement-learning, robotics, tabular-classification, tabular-regression, tabular-to-text, table-to-text, multiple-choice, text-ranking, text-retrieval, time-series-forecasting, text-to-video, image-text-to-text, image-text-to-image, image-text-to-video, visual-question-answering, document-question-answering, zero-shot-image-classification, graph-ml, mask-generation, zero-shot-object-detection, text-to-3d, image-to-3d, image-feature-extraction, video-text-to-text, keypoint-detection, visual-document-retrieval, any-to-any, video-to-video, other

ASL 84 Classes 408D Final Train-Ready Dataset

This dataset contains final train-ready MediaPipe Holistic keypoint sequences for isolated American Sign Language classification.

What was done

  • Kept only classes with at least 50 training samples.
  • Reduced the dataset to 84 classes.
  • Checked and removed bad / zero-heavy sequences.
  • Normalized keypoints.
  • Added velocity / motion features.
  • Converted input features from 204D to 408D.
  • Remapped labels from 0 to 83.

Files

  • train_features_final.npy
  • train_labels_final.npy
  • val_features_final.npy
  • val_labels_final.npy
  • id_to_label_final.json
  • label_to_id_final.json
  • old_to_new_after_cleaning.json
  • new_to_old_after_cleaning.json
  • final_class_counts.csv
  • final_dataset_metadata.json
  • train_sequence_quality_report.csv
  • val_sequence_quality_report.csv

Training configuration

Use these values in the model pipeline:

SEQ_LEN = 50
INPUT_DIM = 408
NUM_CLASSES = 84

Important note

The labels have been remapped after filtering and cleaning. Use id_to_label_final.json during inference to convert predicted class IDs back to labels.

Metadata 

{
  "input_dir": "/kaggle/working/training_model23_filtered_50",
  "output_dir": "/kaggle/working/training_model23_final_train_ready",
  "original_train_shape": [
    6770,
    50,
    204
  ],
  "original_val_shape": [
    862,
    50,
    204
  ],
  "after_bad_sequence_train_shape": [
    6770,
    50,
    204
  ],
  "after_bad_sequence_val_shape": [
    862,
    50,
    204
  ],
  "final_train_shape": [
    6770,
    50,
    408
  ],
  "final_val_shape": [
    862,
    50,
    408
  ],
  "num_classes": 84,
  "normalize_keypoints": true,
  "velocity_added": true,
  "max_zero_frame_ratio": 0.4,
  "min_valid_frame_ratio": 0.6,
  "train_bad_sequence_report": {
    "name": "train",
    "total_samples": 6770,
    "bad_samples": 0,
    "good_samples": 6770,
    "bad_percent": 0.0,
    "avg_valid_frame_ratio": 1.0,
    "avg_zero_frame_ratio": 0.0,
    "nan_samples": 0,
    "inf_samples": 0
  },
  "val_bad_sequence_report": {
    "name": "val",
    "total_samples": 862,
    "bad_samples": 0,
    "good_samples": 862,
    "bad_percent": 0.0,
    "avg_valid_frame_ratio": 1.0,
    "avg_zero_frame_ratio": 0.0,
    "nan_samples": 0,
    "inf_samples": 0
  }
}
Downloads last month
-
Number of rows:
7,632
Total file size:
623 MB