Please enable JavaScript.
Coggle requires JavaScript to display documents.
Automatic Thai-finger spelling Transcription (Alphabet Level Detection…
Automatic Thai-finger spelling Transcription
Alphabet Level Detection
Frame sign sequence
Sign Label and/or Feature sequence
Sign Sequence Classification and Transcription
Base line appoach
LSTM with sign label sequence
LSTM with sign feature sequence
Alphabet Saperation Stage
Heatmap Approach
Frame distance Appoach
Fram distance with one Threshold
Frame distance with two threshold
Moving average additional
Skip frame additional
Window Frame Smooting
Attention Technique
Bahdanau
Loung
Skin Extraction
25 Thai signs Classification
Thai finger spelling localization and classification under
complex background using a YOLO-based deep learning :star:
Hand Sign Recognition for Thai Finger Spelling:
an Application of Convolution Neural Network :star:
Anomaly Detection
(Non-sign/transistion hand image)
Automatic Hand Sign Recognition: Identify Unusuality Through Latent Cognizance. :star:
Latent cognizance: what machine really learns. :star:
:explode: Latent Cognizance
:explode: OpenMax investigate
:explode: Evaluation/Measurement
:explode: Permutation layer information
:no_entry: Not precise
:no_entry: Need inverstigate
:no_entry: Need find new Evaluation
Fooling image
Sign sequence with still image
Encoder-decoder Transcription
with sequence image
:red_cross: train 2-gram test n-gram problem
LSTM Transcription with sign sequence label
Very large Dataset
:warning: 2-gram 1764 samples (42^2)
:warning: 3 gram 74088 samples (42^3)
:red_cross: 4-gram 3111696 samples (42^4)
CNN Classification
CNN Extraction Feature
Feature Vector
Penultimate Vector
One-hot vector
Sign label + word Embedding