Sample 3378265
Run Inference
Sample 3381121
Run Inference
Sample 50802118
Run Inference
Sample 629983
Run Inference
Select a sample and click "Run Inference" to simulate output.
<style>
.signx-analysis.signx-locked {
opacity: 0.6;
cursor: not-allowed;
}
.signx-analysis.signx-active {
background: #007bff;
color: #fff;
border-color: #007bff;
}
.signx-analysis.signx-loading {
color: #202122;
border-color: #a2a9b1;
}
.signx-active-button {
border-color: #ff0000 !important;
box-shadow: 0 0 0 2px rgba(255, 0, 0, 0.35) !important;
}
.signx-flash {
animation: signx-flash 0.8s ease-out;
}
@keyframes signx-flash {
0% { box-shadow: 0 0 0 0 rgba(0, 123, 255, 0.9); }
50% { box-shadow: 0 0 0 6px rgba(0, 123, 255, 0.4); }
100% { box-shadow: 0 0 0 0 rgba(0, 123, 255, 0); }
}
</style>
Analysis 3378265
Analysis 3381121
Analysis 50802118
Analysis 629983
Feature <-> Gloss Map
Locked
Run inference to unlock analysis ↑
![Feature-to-gloss short alignment]()
Gloss-to-Frames
Locked
Run inference to unlock analysis ↑
![Gloss to frames]()
<script>
(function() {
const output = document.getElementById('signx-output');
if (!output) return;
const outputs = {
'3378265': [
'(SignX) user@host:~/project$ bash inference.sh ./eval/tiny_test_data/videos/3378265.mp4',
'',
'======================================================================',
' Sign Language Recognition - Full Inference Pipeline',
'======================================================================',
'',
' Pipeline: Video -> [Latent Space frozen] -> Features -> [CSLR in Latent] -> Gloss',
' Mode: inference (two-stage execution)',
'',
'======================================================================',
'',
'[Configuration]',
' Input video: static/videos/3378265.mp4',
' Output file: inference_output/inference_output_[mm-dd-xxxx]_3378265.txt',
' Latent Space model: latent/work_dir_pose_latent/best_model.pt',
' CSLR in Latent: checkpoints_pose_latent/best',
'',
'[1/2] Extracting video features with Latent Space...',
' Environment: SignX',
'',
' OK Temporary video list created: /tmp/tmp.signx/video_list.txt',
' Loading Latent Space model...',
' [SLRModel] Pose assist encoder initialized',
' OK Feature extraction complete',
' Number of feature sequences: 1',
' OK Source/target placeholder files ready',
'',
'OK Stage 1 complete: features extracted',
'',
'[2/2] Generating gloss sequence with CSLR in Latent...',
' Environment: SignX',
'',
' Loading CSLR in Latent model...',
' Running translation...',
'',
' OK Inference complete: gloss sequence generated',
'',
'Detected detailed attention analysis, saving...',
' OK Saved analysis to: inference_output/detailed_prediction_[mm-dd-xxxx]/3378265',
'',
'Recognition result (BPE removed):',
'----------------------------------------------------------------------',
' With BPE: #IF I@@ X NOT BOR@@ E IX-1p FUTURE READ IX-1p part:indef',
' Predicted Gloss: #IF IX NOT BORE IX-1p FUTURE READ IX-1p part:indef',
' Ground Truth: #IF IX NOT BORE IX-1p FUTURE READ IX-1p part:indef',
'----------------------------------------------------------------------',
'',
'OK Full pipeline completed (Latent Space -> CSLR in Latent)',
'',
'(SignX) user@host:~/project$'
],
'3381121': [
'(SignX) user@host:~/project$ bash inference.sh ./eval/tiny_test_data/videos/3381121.mp4',
'',
'======================================================================',
' Sign Language Recognition - Full Inference Pipeline',
'======================================================================',
'',
' Pipeline: Video -> [Latent Space frozen] -> Features -> [CSLR in Latent] -> Gloss',
' Mode: inference (two-stage execution)',
'',
'======================================================================',
'',
'[Configuration]',
' Input video: static/videos/3381121.mp4',
' Output file: inference_output/inference_output_[mm-dd-xxxx]_3381121.txt',
' Latent Space model: latent/work_dir_pose_latent/best_model.pt',
' CSLR in Latent: checkpoints_pose_latent/best',
'',
'[1/2] Extracting video features with Latent Space...',
' Environment: SignX',
'',
' OK Temporary video list created: /tmp/tmp.signx/video_list.txt',
' Loading Latent Space model...',
' [SLRModel] Pose assist encoder initialized',
' OK Feature extraction complete',
' Number of feature sequences: 1',
' OK Source/target placeholder files ready',
'',
'OK Stage 1 complete: features extracted',
'',
'[2/2] Generating gloss sequence with CSLR in Latent...',
' Environment: SignX',
'',
' Loading CSLR in Latent model...',
' Running translation...',
'',
' OK Inference complete: gloss sequence generated',
'',
'Detected detailed attention analysis, saving...',
' OK Saved analysis to: inference_output/detailed_prediction_[mm-dd-xxxx]/3381121',
'',
'Recognition result (BPE removed):',
'----------------------------------------------------------------------',
' With BPE: BOX/ROOM I@@ X NOT-YET ARRIVE I@@ X SHOULD CONT@@ ACT ns-fs-@@ F@@ E@@ DE@@ X',
' Predicted Gloss: BOX/ROOM IX NOT-YET ARRIVE IX SHOULD CONTACT ns-fs-FEDEX',
' Ground Truth: BOX/ROOM IX NOT-YET ARRIVE IX SHOULD CONTACT ns-fs-FEDEX',
'----------------------------------------------------------------------',
'',
'OK Full pipeline completed (Latent Space -> CSLR in Latent)',
'',
'(SignX) user@host:~/project$'
],
'50802118': [
'(SignX) user@host:~/project$ bash inference.sh ./eval/tiny_test_data/videos/50802118.mp4',
'',
'======================================================================',
' Sign Language Recognition - Full Inference Pipeline',
'======================================================================',
'',
' Pipeline: Video -> [Latent Space frozen] -> Features -> [CSLR in Latent] -> Gloss',
' Mode: inference (two-stage execution)',
'',
'======================================================================',
'',
'[Configuration]',
' Input video: static/videos/50802118.mp4',
' Output file: inference_output/inference_output_[mm-dd-xxxx]_50802118.txt',
' Latent Space model: latent/work_dir_pose_latent/best_model.pt',
' CSLR in Latent: checkpoints_pose_latent/best',
'',
'[1/2] Extracting video features with Latent Space...',
' Environment: SignX',
'',
' OK Temporary video list created: /tmp/tmp.signx/video_list.txt',
' Loading Latent Space model...',
' [SLRModel] Pose assist encoder initialized',
' OK Feature extraction complete',
' Number of feature sequences: 1',
' OK Source/target placeholder files ready',
'',
'OK Stage 1 complete: features extracted',
'',
'[2/2] Generating gloss sequence with CSLR in Latent...',
' Environment: SignX',
'',
' Loading CSLR in Latent model...',
' Running translation...',
'',
' OK Inference complete: gloss sequence generated',
'',
'Detected detailed attention analysis, saving...',
' OK Saved analysis to: inference_output/detailed_prediction_[mm-dd-xxxx]/50802118',
'',
'Recognition result (BPE removed):',
'----------------------------------------------------------------------',
' With BPE: I@@ X ns-fs-JOHN HAVE DIFFERENT C@@ A@@ R',
' Predicted Gloss: IX ns-fs-JOHN HAVE DIFFERENT CAR',
' Ground Truth: IX ns-fs-JOHN HAVE DIFFERENT CAR',
'----------------------------------------------------------------------',
'',
'OK Full pipeline completed (Latent Space -> CSLR in Latent)',
'',
'(SignX) user@host:~/project$'
],
'629983': [
'(SignX) user@host:~/project$ bash inference.sh ./eval/tiny_test_data/videos/629983.mp4',
'',
'======================================================================',
' Sign Language Recognition - Full Inference Pipeline',
'======================================================================',
'',
' Pipeline: Video -> [Latent Space frozen] -> Features -> [CSLR in Latent] -> Gloss',
' Mode: inference (two-stage execution)',
'',
'======================================================================',
'',
'[Configuration]',
' Input video: static/videos/629983.mp4',
' Output file: inference_output/inference_output_[mm-dd-xxxx]_629983.txt',
' Latent Space model: latent/work_dir_pose_latent/best_model.pt',
' CSLR in Latent: checkpoints_pose_latent/best',
'',
'[1/2] Extracting video features with Latent Space...',
' Environment: SignX',
'',
' OK Temporary video list created: /tmp/tmp.signx/video_list.txt',
' Loading Latent Space model...',
' [SLRModel] Pose assist encoder initialized',
' OK Feature extraction complete',
' Number of feature sequences: 1',
' OK Source/target placeholder files ready',
'',
'OK Stage 1 complete: features extracted',
'',
'[2/2] Generating gloss sequence with CSLR in Latent...',
' Environment: SignX',
'',
' Loading CSLR in Latent model...',
' Running translation...',
'',
' OK Inference complete: gloss sequence generated',
'',
'Detected detailed attention analysis, saving...',
' OK Saved analysis to: inference_output/detailed_prediction_[mm-dd-xxxx]/629983',
'',
'Recognition result (BPE removed):',
'----------------------------------------------------------------------',
' With BPE: I@@ X FINISH WORK WH@@ E@@ N I@@ X',
' Predicted Gloss: IX FINISH WORK WHEN IX',
' Ground Truth: IX FINISH WORK WHEN IX',
'----------------------------------------------------------------------',
'',
'OK Full pipeline completed (Latent Space -> CSLR in Latent)',
'',
'(SignX) user@host:~/project$'
]
};
let typingTimer = null;
function renderLines(lines, onProgress) {
if (typingTimer) {
clearInterval(typingTimer);
typingTimer = null;
}
output.textContent = '';
const batches = [];
for (let i = 0; i < lines.length; i += 1) {
const line = lines[i];
if (line.startsWith('[Configuration]')) {
const group = [line];
while (i + 1 < lines.length && lines[i + 1].startsWith(' ')) {
i += 1;
group.push(lines[i]);
}
batches.push({ lines: group, baseDelay: 90 });
continue;
}
if (line.startsWith('====') || line.startsWith(' Pipeline:') || line.startsWith(' Mode:')) {
const group = [line];
while (i + 1 < lines.length && lines[i + 1].startsWith('====')) {
i += 1;
group.push(lines[i]);
}
batches.push({ lines: group, baseDelay: 60 });
continue;
}
if (
line.includes('Extracting features') ||
line.includes('Processing sample') ||
line.includes('Regenerating') ||
line.includes('Saving') ||
line.includes('Extracting') ||
line.includes('Mapping') ||
line.includes('Running translation')
) {
batches.push({ lines: [line], baseDelay: 240 });
continue;
}
if (line.trim() === '') {
batches.push({ lines: [''], baseDelay: 50 });
continue;
}
batches.push({ lines: [line], baseDelay: 120 });
}
const totalMs = 3000 + Math.floor(Math.random() * 8000);
const baseSum = batches.reduce((sum, batch) => sum + batch.baseDelay, 0);
const scale = baseSum > 0 ? totalMs / baseSum : 1;
let batchIndex = 0;
const emitBatch = () => {
if (batchIndex >= batches.length) {
typingTimer = null;
if (onProgress) {
onProgress(100, true);
}
return;
}
const batch = batches[batchIndex];
output.textContent += batch.lines.join('\n') + '\n';
output.scrollTop = output.scrollHeight;
batchIndex += 1;
if (onProgress) {
const progress = Math.floor((batchIndex / batches.length) * 100);
onProgress(progress, false);
}
const jitter = Math.floor(Math.random() * 30) - 10;
const delay = Math.max(40, Math.floor(batch.baseDelay * scale) + jitter);
typingTimer = setTimeout(emitBatch, delay);
};
emitBatch();
}
function showLogs(sampleId) {
const lines = outputs[sampleId];
if (!lines) {
output.textContent = 'No output available for this sample.';
return;
}
output.textContent = lines.join('\n') + '\n';
}
function flashDemoCard(sampleId) {
const card = document.querySelector(`.demo-item[data-sample="${sampleId}"]`);
if (!card) return;
card.classList.remove('signx-flash');
void card.offsetWidth;
card.classList.add('signx-flash');
}
function flashAnalysisImages() {
const targets = [alignImg, glossImg].filter(Boolean);
targets.forEach((img) => {
img.classList.remove('signx-flash');
void img.offsetWidth;
img.classList.add('signx-flash');
});
}
function getButtonSampleId(button) {
if (button.classList.contains('signx-run')) {
const card = button.closest('.demo-item');
return card ? card.getAttribute('data-sample') : null;
}
return button.getAttribute('data-sample');
}
function setActiveButtons(sampleId) {
document.querySelectorAll('.signx-run, .signx-analysis').forEach((btn) => {
const matches = getButtonSampleId(btn) === sampleId;
btn.classList.toggle('signx-active-button', matches);
});
}
document.querySelectorAll('.demo-item').forEach((card) => {
const button = card.querySelector('.signx-run');
if (!button) return;
button.addEventListener('click', () => {
const sample = card.getAttribute('data-sample');
setActiveButtons(sample);
const lines = outputs[sample];
if (!lines) {
output.textContent = 'No output available for this sample.';
return;
}
startAnalysisProgress(sample);
renderLines(lines, (progress, done) => {
setAnalysisProgress(sample, progress);
if (done) {
finishAnalysisProgress(sample);
setAnalysis(sample);
}
});
output.scrollIntoView({ behavior: 'smooth', block: 'start' });
});
});
const analysisMap = {
'3378265': {
align: 'static/images/3378265_frame_alignment_short.png',
gloss: 'static/images/3378265_gloss_to_frames.png',
text: 'With BPE: #IF I@@ X NOT BOR@@ E IX-1p FUTURE READ IX-1p part:indef\nPredicted Gloss: #IF IX NOT BORE IX-1p FUTURE READ IX-1p part:indef\nGround Truth: #IF IX NOT BORE IX-1p FUTURE READ IX-1p part:indef'
},
'3381121': {
align: 'static/images/3381121_frame_alignment_short.png',
gloss: 'static/images/3381121_gloss_to_frames.png',
text: 'With BPE: BOX/ROOM I@@ X NOT-YET ARRIVE I@@ X SHOULD CONT@@ ACT ns-fs-@@ F@@ E@@ DE@@ X\nPredicted Gloss: BOX/ROOM IX NOT-YET ARRIVE IX SHOULD CONTACT ns-fs-FEDEX\nGround Truth: BOX/ROOM IX NOT-YET ARRIVE IX SHOULD CONTACT ns-fs-FEDEX'
},
'50802118': {
align: 'static/images/50802118_frame_alignment_short.png',
gloss: 'static/images/50802118_gloss_to_frames.png',
text: 'With BPE: I@@ X ns-fs-JOHN HAVE DIFFERENT C@@ A@@ R\nPredicted Gloss: IX ns-fs-JOHN HAVE DIFFERENT CAR\nGround Truth: IX ns-fs-JOHN HAVE DIFFERENT CAR'
},
'629983': {
align: 'static/images/629983_frame_alignment_short.png',
gloss: 'static/images/629983_gloss_to_frames.png',
text: 'With BPE: I@@ X FINISH WORK WH@@ E@@ N I@@ X\nPredicted Gloss: IX FINISH WORK WHEN IX\nGround Truth: IX FINISH WORK WHEN IX'
}
};
const alignImg = document.getElementById('analysis-align');
const glossImg = document.getElementById('analysis-gloss');
const alignPlaceholder = document.getElementById('analysis-align-placeholder');
const glossPlaceholder = document.getElementById('analysis-gloss-placeholder');
const analysisText = document.getElementById('analysis-text');
function setAnalysis(sampleId) {
const entry = analysisMap[sampleId];
if (!entry) return;
alignImg.src = entry.align;
glossImg.src = entry.gloss;
if (analysisText) {
analysisText.textContent = entry.text || '';
}
}
document.querySelectorAll('.signx-analysis').forEach((button) => {
button.addEventListener('click', () => {
if (button.classList.contains('signx-locked')) {
return;
}
if (button.classList.contains('signx-loading')) {
return;
}
const sample = button.getAttribute('data-sample');
setActiveButtons(sample);
setAnalysis(sample);
setActiveAnalysis(sample);
showLogs(sample);
flashDemoCard(sample);
flashAnalysisImages();
});
});
function setActiveAnalysis(sampleId) {
document.querySelectorAll('.signx-analysis').forEach((btn) => {
btn.classList.toggle('signx-active', btn.getAttribute('data-sample') === sampleId);
});
}
function setAnalysisProgress(sampleId, percent) {
const button = document.querySelector(`.signx-analysis[data-sample="${sampleId}"]`);
if (!button) return;
if (percent >= 100) {
button.style.background = '#007bff';
button.style.color = '#fff';
button.style.borderColor = '#007bff';
return;
}
button.style.background = `linear-gradient(90deg, #007bff ${percent}%, #f8f9fa ${percent}%)`;
button.style.color = '#202122';
}
function startAnalysisProgress(sampleId) {
const button = document.querySelector(`.signx-analysis[data-sample="${sampleId}"]`);
if (!button) return;
button.classList.remove('signx-locked');
button.classList.remove('signx-active');
button.classList.add('signx-loading');
const icon = button.querySelector('i');
if (icon) {
icon.classList.remove('fa-lock');
icon.classList.add('fa-lock-open');
}
setAnalysisProgress(sampleId, 0);
setActiveAnalysis(sampleId);
}
function finishAnalysisProgress(sampleId) {
const button = document.querySelector(`.signx-analysis[data-sample="${sampleId}"]`);
if (!button) return;
button.classList.remove('signx-loading');
setActiveAnalysis(sampleId);
if (alignPlaceholder && glossPlaceholder) {
alignPlaceholder.style.display = 'none';
glossPlaceholder.style.display = 'none';
}
if (alignImg && glossImg) {
alignImg.style.display = 'block';
glossImg.style.display = 'block';
}
setAnalysisProgress(sampleId, 100);
}
})();
</script>