A relatively perfect version

SignX/{detailed_prediction_20260101_131106 → detailed_prediction_20260101_132358}/3381121/analysis_report.txt

@@ -2,7 +2,7 @@
   Sign Language Recognition - Attention分析报告
 ================================================================================
 
-生成时间: 2026-01-01 13:11:10
+生成时间: 2026-01-01 13:24:03
 
 翻译结果:
 --------------------------------------------------------------------------------

SignX/{detailed_prediction_20260101_131106 → detailed_prediction_20260101_132358}/3381121/attention_keyframes/keyframes_index.txt

@@ -1,7 +1,7 @@
 关键帧索引
 ============================================================
 
-样本目录: /common/users/sf895/output/huggingface_asllrp_repo/SignX/detailed_prediction_20260101_131106/3381121
+样本目录: /common/users/sf895/output/huggingface_asllrp_repo/SignX/detailed_prediction_20260101_132358/3381121
 视频路径: /common/users/sf895/output/huggingface_asllrp_repo/SignX/eval/tiny_test_data/good_videos/3381121.mp4
 总关键帧数: 30
 

SignX/detailed_prediction_20260101_133848/3381121/analysis_report.txt

@@ -0,0 +1,43 @@
+================================================================================
+  Sign Language Recognition - Attention分析报告
+================================================================================
+
+生成时间: 2026-01-01 13:38:55
+
+翻译结果:
+--------------------------------------------------------------------------------
+BOX/ROOM IX NOT-YET ARRIVE IX SHOULD CONTACT ns-fs-FEDEX
+
+视频信息:
+--------------------------------------------------------------------------------
+总帧数: 35
+词数量: 8
+
+Attention权重信息:
+--------------------------------------------------------------------------------
+形状: (30, 35)
+  - 解码步数: 30
+
+词-帧对应详情:
+================================================================================
+No.   Word                 Frames          Peak     Attn     Conf      
+--------------------------------------------------------------------------------
+1     BOX/ROOM             4-4             4        0.618    high      
+2     IX                   7-7             7        0.524    high      
+3     NOT-YET              7-7             7        0.266    medium    
+4     ARRIVE               8-10            8        0.308    medium    
+5     IX                   11-11           11       0.486    medium    
+6     SHOULD               13-13           13       0.595    high      
+7     CONTACT              13-13           13       0.179    low       
+8     ns-fs-FEDEX          17-17           17       0.761    high      
+
+================================================================================
+
+统计摘要:
+--------------------------------------------------------------------------------
+平均attention权重: 0.467
+高置信度词: 4 (50.0%)
+中置信度词: 3 (37.5%)
+低置信度词: 1 (12.5%)
+
+================================================================================

SignX/detailed_prediction_20260101_133848/3381121/attention_keyframes/keyframes_index.txt

@@ -0,0 +1,39 @@
+关键帧索引
+============================================================
+
+样本目录: /common/users/sf895/output/huggingface_asllrp_repo/SignX/detailed_prediction_20260101_133848/3381121
+视频路径: /common/users/sf895/output/huggingface_asllrp_repo/SignX/eval/tiny_test_data/good_videos/3381121.mp4
+总关键帧数: 30
+
+关键帧列表:
+------------------------------------------------------------
+Gloss   0: keyframe_000_feat4_frame17_att0.618.jpg
+Gloss   1: keyframe_001_feat7_frame29_att0.524.jpg
+Gloss   2: keyframe_002_feat7_frame29_att0.266.jpg
+Gloss   3: keyframe_003_feat8_frame32_att0.316.jpg
+Gloss   4: keyframe_004_feat11_frame44_att0.486.jpg
+Gloss   5: keyframe_005_feat13_frame52_att0.595.jpg
+Gloss   6: keyframe_006_feat13_frame52_att0.179.jpg
+Gloss   7: keyframe_007_feat17_frame67_att0.761.jpg
+Gloss   8: keyframe_008_feat21_frame83_att0.176.jpg
+Gloss   9: keyframe_009_feat22_frame87_att0.085.jpg
+Gloss  10: keyframe_010_feat25_frame99_att0.222.jpg
+Gloss  11: keyframe_011_feat28_frame110_att0.069.jpg
+Gloss  12: keyframe_012_feat32_frame126_att0.146.jpg
+Gloss  13: keyframe_013_feat32_frame126_att0.088.jpg
+Gloss  14: keyframe_014_feat34_frame134_att0.117.jpg
+Gloss  15: keyframe_015_feat32_frame126_att0.153.jpg
+Gloss  16: keyframe_016_feat32_frame126_att0.090.jpg
+Gloss  17: keyframe_017_feat34_frame134_att0.116.jpg
+Gloss  18: keyframe_018_feat34_frame134_att0.119.jpg
+Gloss  19: keyframe_019_feat34_frame134_att0.127.jpg
+Gloss  20: keyframe_020_feat34_frame134_att0.128.jpg
+Gloss  21: keyframe_021_feat32_frame126_att0.105.jpg
+Gloss  22: keyframe_022_feat34_frame134_att0.139.jpg
+Gloss  23: keyframe_023_feat34_frame134_att0.146.jpg
+Gloss  24: keyframe_024_feat34_frame134_att0.149.jpg
+Gloss  25: keyframe_025_feat34_frame134_att0.154.jpg
+Gloss  26: keyframe_026_feat34_frame134_att0.157.jpg
+Gloss  27: keyframe_027_feat34_frame134_att0.115.jpg
+Gloss  28: keyframe_028_feat34_frame134_att0.161.jpg
+Gloss  29: keyframe_029_feat34_frame134_att0.144.jpg

SignX/detailed_prediction_20260101_133848/3381121/attention_weights.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f75d778a2e7e7598b861f3b4b05e86ee83ce116f7594899cdac92ec25de5b2f2
+size 4328

SignX/detailed_prediction_20260101_133848/3381121/debug_video_path.txt

@@ -0,0 +1,4 @@
+video_path = '/common/users/sf895/output/huggingface_asllrp_repo/SignX/eval/tiny_test_data/good_videos/3381121.mp4'
+video_path type = <class 'str'>
+video_path is None: False
+bool(video_path): True

SignX/detailed_prediction_20260101_133848/3381121/feature_frame_mapping.json

@@ -0,0 +1,218 @@
+{
+  "original_frame_count": 136,
+  "feature_count": 35,
+  "downsampling_ratio": 3.8857142857142857,
+  "fps": 30.0,
+  "mapping": [
+    {
+      "feature_index": 0,
+      "frame_start": 0,
+      "frame_end": 3,
+      "frame_count": 3
+    },
+    {
+      "feature_index": 1,
+      "frame_start": 3,
+      "frame_end": 7,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 2,
+      "frame_start": 7,
+      "frame_end": 11,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 3,
+      "frame_start": 11,
+      "frame_end": 15,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 4,
+      "frame_start": 15,
+      "frame_end": 19,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 5,
+      "frame_start": 19,
+      "frame_end": 23,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 6,
+      "frame_start": 23,
+      "frame_end": 27,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 7,
+      "frame_start": 27,
+      "frame_end": 31,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 8,
+      "frame_start": 31,
+      "frame_end": 34,
+      "frame_count": 3
+    },
+    {
+      "feature_index": 9,
+      "frame_start": 34,
+      "frame_end": 38,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 10,
+      "frame_start": 38,
+      "frame_end": 42,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 11,
+      "frame_start": 42,
+      "frame_end": 46,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 12,
+      "frame_start": 46,
+      "frame_end": 50,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 13,
+      "frame_start": 50,
+      "frame_end": 54,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 14,
+      "frame_start": 54,
+      "frame_end": 58,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 15,
+      "frame_start": 58,
+      "frame_end": 62,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 16,
+      "frame_start": 62,
+      "frame_end": 66,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 17,
+      "frame_start": 66,
+      "frame_end": 69,
+      "frame_count": 3
+    },
+    {
+      "feature_index": 18,
+      "frame_start": 69,
+      "frame_end": 73,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 19,
+      "frame_start": 73,
+      "frame_end": 77,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 20,
+      "frame_start": 77,
+      "frame_end": 81,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 21,
+      "frame_start": 81,
+      "frame_end": 85,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 22,
+      "frame_start": 85,
+      "frame_end": 89,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 23,
+      "frame_start": 89,
+      "frame_end": 93,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 24,
+      "frame_start": 93,
+      "frame_end": 97,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 25,
+      "frame_start": 97,
+      "frame_end": 101,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 26,
+      "frame_start": 101,
+      "frame_end": 104,
+      "frame_count": 3
+    },
+    {
+      "feature_index": 27,
+      "frame_start": 104,
+      "frame_end": 108,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 28,
+      "frame_start": 108,
+      "frame_end": 112,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 29,
+      "frame_start": 112,
+      "frame_end": 116,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 30,
+      "frame_start": 116,
+      "frame_end": 120,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 31,
+      "frame_start": 120,
+      "frame_end": 124,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 32,
+      "frame_start": 124,
+      "frame_end": 128,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 33,
+      "frame_start": 128,
+      "frame_end": 132,
+      "frame_count": 4
+    },
+    {
+      "feature_index": 34,
+      "frame_start": 132,
+      "frame_end": 136,
+      "frame_count": 4
+    }
+  ]
+}

SignX/detailed_prediction_20260101_133848/3381121/frame_alignment.json

@@ -0,0 +1,86 @@
+{
+  "translation": "BOX/ROOM IX NOT-YET ARRIVE IX SHOULD CONTACT ns-fs-FEDEX",
+  "words": [
+    "BOX/ROOM",
+    "IX",
+    "NOT-YET",
+    "ARRIVE",
+    "IX",
+    "SHOULD",
+    "CONTACT",
+    "ns-fs-FEDEX"
+  ],
+  "total_video_frames": 35,
+  "frame_ranges": [
+    {
+      "word": "BOX/ROOM",
+      "start_frame": 4,
+      "end_frame": 4,
+      "peak_frame": 4,
+      "avg_attention": 0.6183844208717346,
+      "confidence": "high"
+    },
+    {
+      "word": "IX",
+      "start_frame": 7,
+      "end_frame": 7,
+      "peak_frame": 7,
+      "avg_attention": 0.5238878726959229,
+      "confidence": "high"
+    },
+    {
+      "word": "NOT-YET",
+      "start_frame": 7,
+      "end_frame": 7,
+      "peak_frame": 7,
+      "avg_attention": 0.26562702655792236,
+      "confidence": "medium"
+    },
+    {
+      "word": "ARRIVE",
+      "start_frame": 8,
+      "end_frame": 10,
+      "peak_frame": 8,
+      "avg_attention": 0.3079790771007538,
+      "confidence": "medium"
+    },
+    {
+      "word": "IX",
+      "start_frame": 11,
+      "end_frame": 11,
+      "peak_frame": 11,
+      "avg_attention": 0.48612749576568604,
+      "confidence": "medium"
+    },
+    {
+      "word": "SHOULD",
+      "start_frame": 13,
+      "end_frame": 13,
+      "peak_frame": 13,
+      "avg_attention": 0.5946471691131592,
+      "confidence": "high"
+    },
+    {
+      "word": "CONTACT",
+      "start_frame": 13,
+      "end_frame": 13,
+      "peak_frame": 13,
+      "avg_attention": 0.17889028787612915,
+      "confidence": "low"
+    },
+    {
+      "word": "ns-fs-FEDEX",
+      "start_frame": 17,
+      "end_frame": 17,
+      "peak_frame": 17,
+      "avg_attention": 0.7611479163169861,
+      "confidence": "high"
+    }
+  ],
+  "statistics": {
+    "avg_confidence": 0.46708640828728676,
+    "high_confidence_words": 4,
+    "medium_confidence_words": 3,
+    "low_confidence_words": 1
+  }
+}

SignX/detailed_prediction_20260101_133848/3381121/interactive_alignment.html

@@ -0,0 +1,579 @@
+<!DOCTYPE html>
+<html lang="zh-CN">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>Interactive Word-Frame Alignment</title>
+    <style>
+        body {
+            font-family: 'Arial', sans-serif;
+            margin: 20px;
+            background-color: #f5f5f5;
+        }
+        .container {
+            max-width: 1800px;
+            margin: 0 auto;
+            background-color: white;
+            padding: 30px;
+            border-radius: 8px;
+            box-shadow: 0 2px 10px rgba(0,0,0,0.1);
+        }
+        h1 {
+            color: #333;
+            border-bottom: 3px solid #4CAF50;
+            padding-bottom: 10px;
+            margin-bottom: 20px;
+        }
+        .stats {
+            background-color: #E3F2FD;
+            padding: 15px;
+            border-radius: 5px;
+            margin-bottom: 20px;
+            border-left: 4px solid #2196F3;
+            font-size: 14px;
+        }
+        .controls {
+            background-color: #f9f9f9;
+            padding: 20px;
+            border-radius: 5px;
+            margin-bottom: 30px;
+            border: 1px solid #ddd;
+        }
+        .control-group {
+            margin-bottom: 15px;
+        }
+        label {
+            font-weight: bold;
+            display: inline-block;
+            width: 250px;
+            color: #555;
+        }
+        input[type="range"] {
+            width: 400px;
+            vertical-align: middle;
+        }
+        .value-display {
+            display: inline-block;
+            width: 80px;
+            font-family: monospace;
+            font-size: 14px;
+            color: #2196F3;
+            font-weight: bold;
+        }
+        .reset-btn {
+            margin-top: 15px;
+            padding: 10px 25px;
+            background-color: #2196F3;
+            color: white;
+            border: none;
+            border-radius: 5px;
+            cursor: pointer;
+            font-size: 14px;
+            font-weight: bold;
+        }
+        .reset-btn:hover {
+            background-color: #1976D2;
+        }
+        canvas {
+            border: 1px solid #999;
+            display: block;
+            margin: 20px auto;
+            background: white;
+        }
+        .legend {
+            margin-top: 20px;
+            padding: 15px;
+            background-color: #fff;
+            border: 1px solid #ddd;
+            border-radius: 5px;
+        }
+        .legend-item {
+            display: inline-block;
+            margin-right: 25px;
+            font-size: 13px;
+            margin-bottom: 10px;
+        }
+        .color-box {
+            display: inline-block;
+            width: 30px;
+            height: 15px;
+            margin-right: 8px;
+            vertical-align: middle;
+            border: 1px solid #666;
+        }
+        .info-panel {
+            margin-top: 20px;
+            padding: 15px;
+            background-color: #f9f9f9;
+            border-radius: 5px;
+            border: 1px solid #ddd;
+        }
+        .confidence {
+            display: inline-block;
+            padding: 3px 10px;
+            border-radius: 10px;
+            font-weight: bold;
+            font-size: 11px;
+            text-transform: uppercase;
+        }
+        .confidence.high {
+            background-color: #4CAF50;
+            color: white;
+        }
+        .confidence.medium {
+            background-color: #FF9800;
+            color: white;
+        }
+        .confidence.low {
+            background-color: #f44336;
+            color: white;
+        }
+    </style>
+</head>
+<body>
+    <div class="container">
+        <h1>🎯 Interactive Word-to-Frame Alignment Visualizer</h1>
+
+        <div class="stats">
+            <strong>Translation:</strong> BOX/ROOM IX NOT-YET ARRIVE IX SHOULD CONTACT ns-fs-FEDEX<br>
+            <strong>Total Words:</strong> 8 |
+            <strong>Total Features:</strong> 35
+        </div>
+
+        <div class="controls">
+            <h3>⚙️ Threshold Controls</h3>
+
+            <div class="control-group">
+                <label for="peak-threshold">Peak Threshold (% of max):</label>
+                <input type="range" id="peak-threshold" min="1" max="100" value="90" step="1">
+                <span class="value-display" id="peak-threshold-value">90%</span>
+                <br>
+                <small style="margin-left: 255px; color: #666;">
+                    帧的注意力权重 ≥ (峰值权重 × 阈值%) 时被认为是"显著帧"
+                </small>
+            </div>
+
+            <div class="control-group">
+                <label for="confidence-high">High Confidence (avg attn >):</label>
+                <input type="range" id="confidence-high" min="0" max="100" value="50" step="1">
+                <span class="value-display" id="confidence-high-value">0.50</span>
+            </div>
+
+            <div class="control-group">
+                <label for="confidence-medium">Medium Confidence (avg attn >):</label>
+                <input type="range" id="confidence-medium" min="0" max="100" value="20" step="1">
+                <span class="value-display" id="confidence-medium-value">0.20</span>
+            </div>
+
+            <button class="reset-btn" onclick="resetDefaults()">
+                Reset to Defaults
+            </button>
+        </div>
+
+        <div>
+            <h3>Word-to-Frame Alignment</h3>
+            <p style="color: #666; font-size: 13px;">
+                每个词显示为彩色矩形，宽度表示该词对应的特征帧范围。★ = 峰值帧。矩形内部显示注意力权重波形。
+            </p>
+            <canvas id="alignment-canvas" width="1600" height="600"></canvas>
+
+            <h3 style="margin-top: 30px;">Timeline Progress Bar</h3>
+            <canvas id="timeline-canvas" width="1600" height="100"></canvas>
+
+            <div class="legend">
+                <strong>Legend:</strong><br><br>
+                <div class="legend-item">
+                    <span class="confidence high">High</span>
+                    <span class="confidence medium">Medium</span>
+                    <span class="confidence low">Low</span>
+                    Confidence Levels (opacity reflects confidence)
+                </div>
+                <div class="legend-item">
+                    <span style="color: red; font-size: 20px;">★</span>
+                    Peak Frame (highest attention)
+                </div>
+                <div class="legend-item">
+                    <span style="color: blue;">━</span>
+                    Attention Waveform (within word region)
+                </div>
+            </div>
+        </div>
+
+        <div class="info-panel">
+            <h3>Alignment Details</h3>
+            <div id="alignment-details"></div>
+        </div>
+    </div>
+
+    <script>
+        // Attention data from Python
+        const attentionData = [{"word": "BOX/ROOM", "word_idx": 0, "weights": [0.006351051852107048, 0.006571591831743717, 0.012744346633553505, 0.25818338990211487, 0.6183844208717346, 0.07160329818725586, 0.0038708881475031376, 0.0009234889294020832, 0.006989854387938976, 0.004734584596008062, 0.005883616860955954, 0.0007752194069325924, 0.00019075380987487733, 2.629558821354294e-06, 7.526499302912271e-06, 2.065691842290107e-05, 5.081619747215882e-05, 0.0004794567357748747, 0.0001349998638033867, 8.269475074484944e-05, 0.00010472737631062046, 7.984995318111032e-05, 6.447096529882401e-05, 7.145031850086525e-05, 0.00010799866140587255, 0.00015245474060066044, 0.00019921209604945034, 0.0001767162320902571, 0.00017511387704871595, 0.00020259163284208626, 0.00013320642756298184, 9.409035556018353e-05, 0.00012235736357979476, 0.00015512105892412364, 0.00017536635277792811]}, {"word": "IX", "word_idx": 1, "weights": [0.0011723862262442708, 0.0009910413064062595, 0.0016792321112006903, 0.009498031809926033, 0.014361615292727947, 0.04359763115644455, 0.280988484621048, 0.5238878726959229, 0.05194198712706566, 0.03270686790347099, 0.03140342980623245, 0.003325593890622258, 0.001596319954842329, 0.0011391377774998546, 0.0005484184948727489, 0.00033490045461803675, 7.97669927123934e-05, 0.0003001391014549881, 5.5829652410466224e-05, 7.975448170327581e-06, 8.45913564262446e-06, 1.2661466826102696e-05, 1.240926758327987e-05, 8.69539326231461e-06, 9.886168299999554e-06, 1.1390139661671128e-05, 1.0159355042560492e-05, 9.64598439168185e-06, 8.925362635636702e-06, 9.117472473008092e-06, 1.2588812751346268e-05, 2.6802983484230936e-05, 5.668022276950069e-05, 8.233459811890498e-05, 0.00010355122503824532]}, {"word": "NOT-YET", "word_idx": 2, "weights": [0.16965249180793762, 0.07323458045721054, 0.029197975993156433, 0.0025396100245416164, 0.002445423509925604, 0.009898832067847252, 0.037797823548316956, 0.26562702655792236, 0.009848427027463913, 0.00526211503893137, 0.004250594414770603, 0.0014572322834283113, 0.001674243831075728, 0.05451618880033493, 0.060803089290857315, 0.04630552604794502, 0.01521533913910389, 0.004090897738933563, 0.003519815392792225, 0.0038278333377093077, 0.0024895716924220324, 0.002131127519533038, 0.0033031043130904436, 0.0035419934429228306, 0.0020546577870845795, 0.0016667826566845179, 0.0013373601250350475, 0.0012249398278072476, 0.001324663171544671, 0.0019041926134377718, 0.004013519734144211, 0.019279679283499718, 0.04461098462343216, 0.053339678794145584, 0.05661269277334213]}, {"word": "ARRIVE", "word_idx": 3, "weights": [0.0002905388828366995, 0.0002083813596982509, 0.00032632541842758656, 0.0025478217285126448, 0.0073820799589157104, 0.014858342707157135, 0.018400374799966812, 0.021123293787240982, 0.3158378005027771, 0.25220221281051636, 0.30012035369873047, 0.010648821480572224, 0.001886643934994936, 6.4878404373303056e-06, 5.923872322455281e-06, 2.3099497411749326e-05, 0.0005105354939587414, 0.04699746519327164, 0.005902951583266258, 0.00029092541080899537, 0.00018096565327141434, 6.506430509034544e-05, 3.2437703339383006e-05, 2.104153281834442e-05, 7.040531727398047e-06, 7.098288733686786e-06, 7.855384865251835e-06, 6.147487965790788e-06, 5.554756626224844e-06, 7.759556865494233e-06, 6.213985670910915e-06, 1.072721897799056e-05, 1.9002200133400038e-05, 2.5421264581382275e-05, 2.727148421399761e-05]}, {"word": "IX", "word_idx": 4, "weights": [0.0003390431229490787, 0.00022571485897060484, 0.0002557964762672782, 0.0005381361697800457, 0.002120023826137185, 0.0273550096899271, 0.012844335287809372, 0.00290689617395401, 0.01703452318906784, 0.02637772634625435, 0.15619716048240662, 0.48612749576568604, 0.2603294849395752, 0.0005641445168294013, 0.00019721912394743413, 0.00033904434530995786, 0.0006775215733796358, 0.0013669012114405632, 0.0016405221540480852, 0.0007656495436094701, 0.0005228326190263033, 0.0003606425889302045, 0.00021389758330769837, 0.00013279783888719976, 3.7730329495389014e-05, 3.255438059568405e-05, 3.4443997719790787e-05, 3.628108970588073e-05, 3.273988113505766e-05, 3.9840597310103476e-05, 4.062296648044139e-05, 5.8705947594717145e-05, 7.590500899823382e-05, 9.174603474093601e-05, 8.684050408191979e-05]}, {"word": "SHOULD", "word_idx": 5, "weights": [0.00527340080589056, 0.004166516475379467, 0.003337869420647621, 0.0011889089364558458, 0.0007747402414679527, 0.01280419435352087, 0.03180841729044914, 0.04200495034456253, 0.002391757909208536, 0.004637685138732195, 0.006087929010391235, 0.01985923945903778, 0.060313232243061066, 0.5946471691131592, 0.15804460644721985, 0.029653724282979965, 0.0008350771386176348, 0.00013930512068327516, 0.00016234509530477226, 0.00027477910043671727, 0.0006075621349737048, 0.0018652487779036164, 0.0021796557120978832, 0.0014713223790749907, 0.0024460090789943933, 0.002028325106948614, 0.0012717237696051598, 0.001144407782703638, 0.0009023174061439931, 0.0005251378170214593, 0.0008574927924200892, 0.001382339047268033, 0.001502902596257627, 0.001733113662339747, 0.001676460960879922]}, {"word": "CONTACT", "word_idx": 6, "weights": [0.14272911846637726, 0.0573246069252491, 0.0188103336840868, 0.0006784269353374839, 0.00042378040961921215, 0.0014155198587104678, 0.0069302283227443695, 0.08298847824335098, 0.002187008038163185, 0.0017832987941801548, 0.0012008043704554439, 0.0009829651098698378, 0.0021144067868590355, 0.17889028787612915, 0.15413032472133636, 0.09035450220108032, 0.021829815581440926, 0.004844403825700283, 0.006283571477979422, 0.00830614659935236, 0.007968198508024216, 0.008506102487444878, 0.010532977990806103, 0.01149566750973463, 0.008719464763998985, 0.006225862540304661, 0.004270944744348526, 0.003964268136769533, 0.0038093950133770704, 0.004123717080801725, 0.006864133290946484, 0.021106082946062088, 0.03659150376915932, 0.041107941418886185, 0.04050571098923683]}, {"word": "ns-fs-FEDEX", "word_idx": 7, "weights": [7.23023695172742e-05, 5.213047916186042e-05, 4.6853372623445466e-05, 8.384212560486048e-05, 5.267366213956848e-05, 4.4198710384080186e-05, 6.236167973838747e-05, 0.0005651791580021381, 0.004614907316863537, 0.0062443651258945465, 0.00276308530010283, 0.0010768879437819123, 0.0007893505389802158, 0.0001608922757441178, 0.00028600034420378506, 0.0006088865338824689, 0.010051908902823925, 0.7611479163169861, 0.1628209799528122, 0.011173587292432785, 0.01768321916460991, 0.008066349662840366, 0.003297096583992243, 0.002270511817187071, 0.0019132717279717326, 0.0015967994695529342, 0.001093801110982895, 0.0005375861073844135, 0.00038413898437283933, 0.0001939320209203288, 7.028852996882051e-05, 4.9711332394508645e-05, 4.136270945309661e-05, 4.271505167707801e-05, 4.087587149115279e-05]}];
+        const numGlosses = 8;
+        const numFeatures = 35;
+
+        // Colors for different words (matching matplotlib tab20)
+        const colors = [
+            '#1f77b4', '#ff7f0e', '#2ca02c', '#d62728', '#9467bd',
+            '#8c564b', '#e377c2', '#7f7f7f', '#bcbd22', '#17becf',
+            '#aec7e8', '#ffbb78', '#98df8a', '#ff9896', '#c5b0d5',
+            '#c49c94', '#f7b6d2', '#c7c7c7', '#dbdb8d', '#9edae5'
+        ];
+
+        // Get controls
+        const peakThresholdSlider = document.getElementById('peak-threshold');
+        const peakThresholdValue = document.getElementById('peak-threshold-value');
+        const confidenceHighSlider = document.getElementById('confidence-high');
+        const confidenceHighValue = document.getElementById('confidence-high-value');
+        const confidenceMediumSlider = document.getElementById('confidence-medium');
+        const confidenceMediumValue = document.getElementById('confidence-medium-value');
+        const alignmentCanvas = document.getElementById('alignment-canvas');
+        const timelineCanvas = document.getElementById('timeline-canvas');
+        const alignmentCtx = alignmentCanvas.getContext('2d');
+        const timelineCtx = timelineCanvas.getContext('2d');
+
+        // Update displays when sliders change
+        peakThresholdSlider.oninput = function() {
+            peakThresholdValue.textContent = this.value + '%';
+            updateVisualization();
+        };
+
+        confidenceHighSlider.oninput = function() {
+            confidenceHighValue.textContent = (this.value / 100).toFixed(2);
+            updateVisualization();
+        };
+
+        confidenceMediumSlider.oninput = function() {
+            confidenceMediumValue.textContent = (this.value / 100).toFixed(2);
+            updateVisualization();
+        };
+
+        function resetDefaults() {
+            peakThresholdSlider.value = 90;
+            confidenceHighSlider.value = 50;
+            confidenceMediumSlider.value = 20;
+            peakThresholdValue.textContent = '90%';
+            confidenceHighValue.textContent = '0.50';
+            confidenceMediumValue.textContent = '0.20';
+            updateVisualization();
+        }
+
+        function calculateAlignment(weights, peakThreshold) {
+            // Find peak
+            let peakIdx = 0;
+            let peakWeight = weights[0];
+            for (let i = 1; i < weights.length; i++) {
+                if (weights[i] > peakWeight) {
+                    peakWeight = weights[i];
+                    peakIdx = i;
+                }
+            }
+
+            // Find significant frames
+            const threshold = peakWeight * (peakThreshold / 100);
+            let startIdx = peakIdx;
+            let endIdx = peakIdx;
+            let sumWeight = 0;
+            let count = 0;
+
+            for (let i = 0; i < weights.length; i++) {
+                if (weights[i] >= threshold) {
+                    if (i < startIdx) startIdx = i;
+                    if (i > endIdx) endIdx = i;
+                    sumWeight += weights[i];
+                    count++;
+                }
+            }
+
+            const avgWeight = count > 0 ? sumWeight / count : peakWeight;
+
+            return {
+                startIdx: startIdx,
+                endIdx: endIdx,
+                peakIdx: peakIdx,
+                peakWeight: peakWeight,
+                avgWeight: avgWeight,
+                threshold: threshold
+            };
+        }
+
+        function getConfidenceLevel(avgWeight, highThreshold, mediumThreshold) {
+            if (avgWeight > highThreshold) return 'high';
+            if (avgWeight > mediumThreshold) return 'medium';
+            return 'low';
+        }
+
+        function drawAlignmentChart() {
+            const peakThreshold = parseInt(peakThresholdSlider.value);
+            const highThreshold = parseInt(confidenceHighSlider.value) / 100;
+            const mediumThreshold = parseInt(confidenceMediumSlider.value) / 100;
+
+            // Canvas dimensions
+            const width = alignmentCanvas.width;
+            const height = alignmentCanvas.height;
+            const leftMargin = 180;
+            const rightMargin = 50;
+            const topMargin = 60;
+            const bottomMargin = 80;
+
+            const plotWidth = width - leftMargin - rightMargin;
+            const plotHeight = height - topMargin - bottomMargin;
+
+            const rowHeight = plotHeight / numGlosses;
+            const featureWidth = plotWidth / numFeatures;
+
+            // Clear canvas
+            alignmentCtx.clearRect(0, 0, width, height);
+
+            // Draw title
+            alignmentCtx.fillStyle = '#333';
+            alignmentCtx.font = 'bold 18px Arial';
+            alignmentCtx.textAlign = 'center';
+            alignmentCtx.fillText('Word-to-Frame Alignment', width / 2, 30);
+            alignmentCtx.font = '13px Arial';
+            alignmentCtx.fillText('(based on attention peaks, ★ = peak frame)', width / 2, 48);
+
+            // Calculate alignments
+            const alignments = [];
+            for (let wordIdx = 0; wordIdx < numGlosses; wordIdx++) {
+                const data = attentionData[wordIdx];
+                const alignment = calculateAlignment(data.weights, peakThreshold);
+                alignment.word = data.word;
+                alignment.wordIdx = wordIdx;
+                alignment.weights = data.weights;
+                alignments.push(alignment);
+            }
+
+            // Draw grid
+            alignmentCtx.strokeStyle = '#e0e0e0';
+            alignmentCtx.lineWidth = 0.5;
+            for (let i = 0; i <= numFeatures; i++) {
+                const x = leftMargin + i * featureWidth;
+                alignmentCtx.beginPath();
+                alignmentCtx.moveTo(x, topMargin);
+                alignmentCtx.lineTo(x, topMargin + plotHeight);
+                alignmentCtx.stroke();
+            }
+
+            // Draw word regions
+            for (let wordIdx = 0; wordIdx < numGlosses; wordIdx++) {
+                const alignment = alignments[wordIdx];
+                const confidence = getConfidenceLevel(alignment.avgWeight, highThreshold, mediumThreshold);
+                const y = topMargin + wordIdx * rowHeight;
+
+                // Alpha based on confidence
+                const alpha = confidence === 'high' ? 0.9 : confidence === 'medium' ? 0.7 : 0.5;
+
+                // Draw rectangle for word region
+                const startX = leftMargin + alignment.startIdx * featureWidth;
+                const rectWidth = (alignment.endIdx - alignment.startIdx + 1) * featureWidth;
+
+                alignmentCtx.fillStyle = colors[wordIdx % 20];
+                alignmentCtx.globalAlpha = alpha;
+                alignmentCtx.fillRect(startX, y, rectWidth, rowHeight * 0.8);
+                alignmentCtx.globalAlpha = 1.0;
+
+                // Draw border
+                alignmentCtx.strokeStyle = '#000';
+                alignmentCtx.lineWidth = 2;
+                alignmentCtx.strokeRect(startX, y, rectWidth, rowHeight * 0.8);
+
+                // Draw attention waveform inside rectangle
+                alignmentCtx.strokeStyle = 'rgba(0, 0, 255, 0.8)';
+                alignmentCtx.lineWidth = 1.5;
+                alignmentCtx.beginPath();
+                for (let i = alignment.startIdx; i <= alignment.endIdx; i++) {
+                    const x = leftMargin + i * featureWidth + featureWidth / 2;
+                    const weight = alignment.weights[i];
+                    const maxWeight = alignment.peakWeight;
+                    const normalizedWeight = weight / (maxWeight * 1.2); // Scale for visibility
+                    const waveY = y + rowHeight * 0.8 - (normalizedWeight * rowHeight * 0.6);
+
+                    if (i === alignment.startIdx) {
+                        alignmentCtx.moveTo(x, waveY);
+                    } else {
+                        alignmentCtx.lineTo(x, waveY);
+                    }
+                }
+                alignmentCtx.stroke();
+
+                // Draw word label
+                const labelX = startX + rectWidth / 2;
+                const labelY = y + rowHeight * 0.4;
+
+                alignmentCtx.fillStyle = 'rgba(0, 0, 0, 0.7)';
+                alignmentCtx.fillRect(labelX - 60, labelY - 12, 120, 24);
+                alignmentCtx.fillStyle = '#fff';
+                alignmentCtx.font = 'bold 13px Arial';
+                alignmentCtx.textAlign = 'center';
+                alignmentCtx.textBaseline = 'middle';
+                alignmentCtx.fillText(alignment.word, labelX, labelY);
+
+                // Mark peak frame with star
+                const peakX = leftMargin + alignment.peakIdx * featureWidth + featureWidth / 2;
+                const peakY = y + rowHeight * 0.4;
+
+                // Draw star
+                alignmentCtx.fillStyle = '#ff0000';
+                alignmentCtx.strokeStyle = '#ffff00';
+                alignmentCtx.lineWidth = 1.5;
+                alignmentCtx.font = '20px Arial';
+                alignmentCtx.textAlign = 'center';
+                alignmentCtx.strokeText('★', peakX, peakY);
+                alignmentCtx.fillText('★', peakX, peakY);
+
+                // Y-axis label (word names)
+                alignmentCtx.fillStyle = '#333';
+                alignmentCtx.font = '12px Arial';
+                alignmentCtx.textAlign = 'right';
+                alignmentCtx.textBaseline = 'middle';
+                alignmentCtx.fillText(alignment.word, leftMargin - 10, y + rowHeight * 0.4);
+            }
+
+            // Draw horizontal grid lines
+            alignmentCtx.strokeStyle = '#ccc';
+            alignmentCtx.lineWidth = 0.5;
+            for (let i = 0; i <= numGlosses; i++) {
+                const y = topMargin + i * rowHeight;
+                alignmentCtx.beginPath();
+                alignmentCtx.moveTo(leftMargin, y);
+                alignmentCtx.lineTo(leftMargin + plotWidth, y);
+                alignmentCtx.stroke();
+            }
+
+            // Draw axes
+            alignmentCtx.strokeStyle = '#000';
+            alignmentCtx.lineWidth = 2;
+            alignmentCtx.strokeRect(leftMargin, topMargin, plotWidth, plotHeight);
+
+            // X-axis labels (frame indices)
+            alignmentCtx.fillStyle = '#000';
+            alignmentCtx.font = '11px Arial';
+            alignmentCtx.textAlign = 'center';
+            alignmentCtx.textBaseline = 'top';
+            for (let i = 0; i < numFeatures; i++) {
+                const x = leftMargin + i * featureWidth + featureWidth / 2;
+                alignmentCtx.fillText(i.toString(), x, topMargin + plotHeight + 10);
+            }
+
+            // Axis titles
+            alignmentCtx.fillStyle = '#333';
+            alignmentCtx.font = 'bold 14px Arial';
+            alignmentCtx.textAlign = 'center';
+            alignmentCtx.fillText('Feature Frame Index', leftMargin + plotWidth / 2, height - 20);
+
+            alignmentCtx.save();
+            alignmentCtx.translate(30, topMargin + plotHeight / 2);
+            alignmentCtx.rotate(-Math.PI / 2);
+            alignmentCtx.fillText('Generated Word', 0, 0);
+            alignmentCtx.restore();
+
+            return alignments;
+        }
+
+        function drawTimeline(alignments) {
+            const highThreshold = parseInt(confidenceHighSlider.value) / 100;
+            const mediumThreshold = parseInt(confidenceMediumSlider.value) / 100;
+
+            const width = timelineCanvas.width;
+            const height = timelineCanvas.height;
+            const leftMargin = 180;
+            const rightMargin = 50;
+            const plotWidth = width - leftMargin - rightMargin;
+            const featureWidth = plotWidth / numFeatures;
+
+            // Clear canvas
+            timelineCtx.clearRect(0, 0, width, height);
+
+            // Background bar
+            timelineCtx.fillStyle = '#ddd';
+            timelineCtx.fillRect(leftMargin, 30, plotWidth, 40);
+            timelineCtx.strokeStyle = '#000';
+            timelineCtx.lineWidth = 2;
+            timelineCtx.strokeRect(leftMargin, 30, plotWidth, 40);
+
+            // Draw word regions on timeline
+            for (let wordIdx = 0; wordIdx < alignments.length; wordIdx++) {
+                const alignment = alignments[wordIdx];
+                const confidence = getConfidenceLevel(alignment.avgWeight, highThreshold, mediumThreshold);
+                const alpha = confidence === 'high' ? 0.9 : confidence === 'medium' ? 0.7 : 0.5;
+
+                const startX = leftMargin + alignment.startIdx * featureWidth;
+                const rectWidth = (alignment.endIdx - alignment.startIdx + 1) * featureWidth;
+
+                timelineCtx.fillStyle = colors[wordIdx % 20];
+                timelineCtx.globalAlpha = alpha;
+                timelineCtx.fillRect(startX, 30, rectWidth, 40);
+                timelineCtx.globalAlpha = 1.0;
+                timelineCtx.strokeStyle = '#000';
+                timelineCtx.lineWidth = 0.5;
+                timelineCtx.strokeRect(startX, 30, rectWidth, 40);
+            }
+
+            // Title
+            timelineCtx.fillStyle = '#333';
+            timelineCtx.font = 'bold 13px Arial';
+            timelineCtx.textAlign = 'left';
+            timelineCtx.fillText('Timeline Progress Bar', leftMargin, 20);
+        }
+
+        function updateDetailsPanel(alignments, highThreshold, mediumThreshold) {
+            const panel = document.getElementById('alignment-details');
+            let html = '<table style="width: 100%; border-collapse: collapse;">';
+            html += '<tr style="background: #f0f0f0; font-weight: bold;">';
+            html += '<th style="padding: 8px; border: 1px solid #ddd;">Word</th>';
+            html += '<th style="padding: 8px; border: 1px solid #ddd;">Feature Range</th>';
+            html += '<th style="padding: 8px; border: 1px solid #ddd;">Peak</th>';
+            html += '<th style="padding: 8px; border: 1px solid #ddd;">Span</th>';
+            html += '<th style="padding: 8px; border: 1px solid #ddd;">Avg Attention</th>';
+            html += '<th style="padding: 8px; border: 1px solid #ddd;">Confidence</th>';
+            html += '</tr>';
+
+            for (const align of alignments) {
+                const confidence = getConfidenceLevel(align.avgWeight, highThreshold, mediumThreshold);
+                const span = align.endIdx - align.startIdx + 1;
+
+                html += '<tr>';
+                html += `<td style="padding: 8px; border: 1px solid #ddd;"><strong>${align.word}</strong></td>`;
+                html += `<td style="padding: 8px; border: 1px solid #ddd;">${align.startIdx} → ${align.endIdx}</td>`;
+                html += `<td style="padding: 8px; border: 1px solid #ddd;">${align.peakIdx}</td>`;
+                html += `<td style="padding: 8px; border: 1px solid #ddd;">${span}</td>`;
+                html += `<td style="padding: 8px; border: 1px solid #ddd;">${align.avgWeight.toFixed(4)}</td>`;
+                html += `<td style="padding: 8px; border: 1px solid #ddd;"><span class="confidence ${confidence}">${confidence}</span></td>`;
+                html += '</tr>';
+            }
+
+            html += '</table>';
+            panel.innerHTML = html;
+        }
+
+        function updateVisualization() {
+            const alignments = drawAlignmentChart();
+            drawTimeline(alignments);
+            const highThreshold = parseInt(confidenceHighSlider.value) / 100;
+            const mediumThreshold = parseInt(confidenceMediumSlider.value) / 100;
+            updateDetailsPanel(alignments, highThreshold, mediumThreshold);
+        }
+
+        // Event listeners for sliders
+        peakSlider.addEventListener('input', function() {
+            peakValue.textContent = peakSlider.value + '%';
+            updateVisualization();
+        });
+
+        confidenceHighSlider.addEventListener('input', function() {
+            const val = parseInt(confidenceHighSlider.value) / 100;
+            confidenceHighValue.textContent = val.toFixed(2);
+            updateVisualization();
+        });
+
+        confidenceMediumSlider.addEventListener('input', function() {
+            const val = parseInt(confidenceMediumSlider.value) / 100;
+            confidenceMediumValue.textContent = val.toFixed(2);
+            updateVisualization();
+        });
+
+        // Initial visualization
+        updateVisualization();
+    </script>
+</body>
+</html>

SignX/detailed_prediction_20260101_133848/3381121/translation.txt

@@ -0,0 +1,3 @@
+With BPE: BOX/ROOM I@@ X NOT-YET ARRIVE I@@ X SHOULD CONT@@ ACT ns-fs-@@ F@@ E@@ DE@@ X
+Clean: 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

SignX/eval/attention_analysis.py

@@ -256,9 +256,11 @@ class AttentionAnalyzer:
             print("  跳过对齐图: matplotlib未安装")
             return
 
+        output_path = Path(output_path)
+
         # Try to load feature-to-frame mapping
         feature_mapping = None
-        output_dir = Path(output_path).parent
+        output_dir = output_path.parent
         mapping_file = output_dir / "feature_frame_mapping.json"
         if mapping_file.exists():
             try:
@@ -267,152 +269,164 @@ class AttentionAnalyzer:
             except Exception as e:
                 print(f"  Warning: Failed to load feature mapping: {e}")
 
-        # Adjust layout based on whether we have feature mapping
-        if feature_mapping:
-            fig = plt.figure(figsize=(18, 10))
-            gs = GridSpec(4, 1, height_ratios=[4, 1, 1, 0.5], hspace=0.4)
+        if self.word_frame_ranges:
+            max_feat_end = max(w['end_frame'] for w in self.word_frame_ranges)
         else:
-            fig = plt.figure(figsize=(18, 8))
-            gs = GridSpec(3, 1, height_ratios=[4, 1, 0.5], hspace=0.4)
-
-        # === 上图: 词-帧对齐 ===
-        ax1 = fig.add_subplot(gs[0])
-
-        colors = plt.cm.tab20(np.linspace(0, 1, max(len(self.words), 20)))
-
-        for i, word_info in enumerate(self.word_frame_ranges):
-            start = word_info['start_frame']
-            end = word_info['end_frame']
-            word = word_info['word']
-            confidence = word_info['confidence']
-
-            # 根据置信度设置透明度
-            alpha = 0.9 if confidence == 'high' else 0.7 if confidence == 'medium' else 0.5
-
-            # 绘制矩形
-            rect = patches.Rectangle(
-                (start, i), end - start + 1, 0.8,
-                linewidth=2, edgecolor='black',
-                facecolor=colors[i % 20], alpha=alpha
-            )
-            ax1.add_patch(rect)
-
-            # 添加词标签
-            ax1.text(start + (end - start) / 2, i + 0.4, word,
-                    ha='center', va='center', fontsize=11,
-                    fontweight='bold', color='white',
-                    bbox=dict(boxstyle='round,pad=0.3', facecolor='black', alpha=0.5))
-
-            # 标记峰值帧
-            peak = word_info['peak_frame']
-            ax1.plot(peak, i + 0.4, 'r*', markersize=15, markeredgecolor='yellow',
-                    markeredgewidth=1.5)
-
-        ax1.set_xlim(-2, self.video_frames + 2)
-        ax1.set_ylim(-0.5, len(self.words))
-        # Remove redundant label (timeline info shown below)
-        ax1.set_xlabel('')
-        ax1.set_ylabel('Generated Word', fontsize=13, fontweight='bold')
-        ax1.set_title('Word-to-Frame Alignment\n(based on attention peaks, ★ = peak frame)',
-                     fontsize=15, pad=15, fontweight='bold')
-        ax1.grid(True, alpha=0.3, axis='x', linestyle='--')
-        ax1.set_yticks(range(len(self.words)))
-        ax1.set_yticklabels([w['word'] for w in self.word_frame_ranges], fontsize=10)
-        ax1_label_pos = ax1.yaxis.label.get_position()
-
-        # === 中图1: SMKD特征帧时间线进度条 ===
-        ax2 = fig.add_subplot(gs[1])
-
-        # 背景
-        ax2.barh(0, self.video_frames, height=0.6, color='lightgray',
-                edgecolor='black', linewidth=2)
-
-        # 每个词的区域
-        for i, word_info in enumerate(self.word_frame_ranges):
-            start = word_info['start_frame']
-            end = word_info['end_frame']
-            confidence = word_info['confidence']
-            alpha = 0.9 if confidence == 'high' else 0.7 if confidence == 'medium' else 0.5
-
-            ax2.barh(0, end - start + 1, left=start, height=0.6,
-                    color=colors[i % 20], alpha=alpha, edgecolor='black', linewidth=0.5)
-
-        ax2.set_xlim(-2, self.video_frames + 2)
-        ax2.set_ylim(-0.4, 0.4)
-        ax2.set_xlabel('')
-        ax2.set_yticks([])
-        ax2.set_title('Latent Feature Timeline', fontsize=13, fontweight='bold')
-        ax2.grid(True, alpha=0.3, axis='x', linestyle='--')
-
-        # === 中图2: 原始视频帧时间线进度条 (如果有feature mapping) ===
-        timeline_axes = [ax2]
-
+            max_feat_end = self.video_frames - 1
+        latent_full_limit = self.video_frames + 2
+        latent_short_limit = max(min(latent_full_limit, max_feat_end + 2), 5)
+
+        original_frame_count = None
+        mapping_list = None
+        orig_full_limit = None
+        orig_short_limit = None
         if feature_mapping:
-            ax3 = fig.add_subplot(gs[2])
-
-            original_frame_count = feature_mapping['original_frame_count']
+            original_frame_count = feature_mapping.get('original_frame_count', self.video_frames)
+            mapping_list = feature_mapping.get('mapping', [])
+            orig_full_limit = original_frame_count + 2
+            if mapping_list:
+                idx = min(max_feat_end, len(mapping_list) - 1)
+                orig_short_limit = mapping_list[idx]['frame_end'] + 2
+
+        def render_alignment(out_path, latent_xlim_end, orig_xlim_end=None):
+            if feature_mapping:
+                fig = plt.figure(figsize=(18, 10))
+                gs = GridSpec(4, 1, height_ratios=[4, 1, 1, 0.5], hspace=0.4)
+            else:
+                fig = plt.figure(figsize=(18, 8))
+                gs = GridSpec(3, 1, height_ratios=[4, 1, 0.5], hspace=0.4)
 
-            # 背景
-            ax3.barh(0, original_frame_count, height=0.6, color='lightgray',
-                    edgecolor='black', linewidth=2)
+            # === 上图: 词-帧对齐 ===
+            ax1 = fig.add_subplot(gs[0])
+            colors = plt.cm.tab20(np.linspace(0, 1, max(len(self.words), 20)))
 
-            # 每个词对应的原始帧区域
             for i, word_info in enumerate(self.word_frame_ranges):
-                feat_start = word_info['start_frame']
-                feat_end = word_info['end_frame']
+                start = word_info['start_frame']
+                end = word_info['end_frame']
+                word = word_info['word']
                 confidence = word_info['confidence']
                 alpha = 0.9 if confidence == 'high' else 0.7 if confidence == 'medium' else 0.5
 
-                # 从feature mapping中找到对应的原始帧范围
-                # 使用特征帧的起始和结束索引来查找原始帧范围
-                mapping_list = feature_mapping['mapping']
-                if feat_start < len(mapping_list) and feat_end < len(mapping_list):
-                    orig_start = mapping_list[feat_start]['frame_start']
-                    orig_end = mapping_list[feat_end]['frame_end']
-
-                    ax3.barh(0, orig_end - orig_start, left=orig_start, height=0.6,
-                            color=colors[i % 20], alpha=alpha, edgecolor='black', linewidth=0.5)
-
-            ax3.set_xlim(-2, original_frame_count + 2)
-            ax3.set_ylim(-0.4, 0.4)
-            ax3.set_xlabel('')
-            ax3.set_yticks([])
-            ax3.set_title(f'Original Video Timeline ({original_frame_count} frames, '
-                         f'{feature_mapping["downsampling_ratio"]:.2f}x downsampling)',
-                         fontsize=13, fontweight='bold')
-            ax3.grid(True, alpha=0.3, axis='x', linestyle='--')
-
-            legend_row = 3
-            timeline_axes.append(ax3)
-        else:
-            legend_row = 2
-
-        # === 下图: 置信度图例 ===
-        ax_legend = fig.add_subplot(gs[legend_row])
-        ax_legend.axis('off')
-
-        legend_text = "Confidence:  ■ High (avg attn > 0.5)  ■ Medium (0.2-0.5)  ■ Low (< 0.2)"
-        ax_legend.text(0.5, 0.5, legend_text, ha='center', va='center',
-                fontsize=11, transform=ax_legend.transAxes)
-
-        plt.tight_layout()
-        fig.canvas.draw()
+                rect = patches.Rectangle(
+                    (start, i), end - start + 1, 0.8,
+                    linewidth=2, edgecolor='black',
+                    facecolor=colors[i % 20], alpha=alpha
+                )
+                ax1.add_patch(rect)
+
+                ax1.text(start + (end - start) / 2, i + 0.4, word,
+                        ha='center', va='center', fontsize=11,
+                        fontweight='bold', color='white',
+                        bbox=dict(boxstyle='round,pad=0.3', facecolor='black', alpha=0.5))
+
+                peak = word_info['peak_frame']
+                ax1.plot(peak, i + 0.4, 'r*', markersize=15, markeredgecolor='yellow',
+                        markeredgewidth=1.5)
+
+            ax1.set_xlim(-2, latent_xlim_end)
+            ax1.set_ylim(-0.5, len(self.words))
+            ax1.set_xlabel('')
+            ax1.set_ylabel('')
+            ax1.set_title('Word-to-Frame Alignment\n(based on attention peaks, ★ = peak frame)',
+                         fontsize=15, pad=15, fontweight='bold')
+            ax1.grid(True, alpha=0.3, axis='x', linestyle='--')
+            ax1.set_yticks(range(len(self.words)))
+            ax1.set_yticklabels([w['word'] for w in self.word_frame_ranges], fontsize=10)
+
+            # === 中图1: Latent timeline ===
+            ax2 = fig.add_subplot(gs[1])
+            ax2.barh(0, self.video_frames, height=0.6, color='lightgray',
+                    edgecolor='black', linewidth=2)
+            for i, word_info in enumerate(self.word_frame_ranges):
+                start = word_info['start_frame']
+                end = word_info['end_frame']
+                confidence = word_info['confidence']
+                alpha = 0.9 if confidence == 'high' else 0.7 if confidence == 'medium' else 0.5
+                ax2.barh(0, end - start + 1, left=start, height=0.6,
+                        color=colors[i % 20], alpha=alpha, edgecolor='black', linewidth=0.5)
 
-        # Draw shared Timeline label aligned with Generated Word label (after layout)
-        label_disp = ax1.transAxes.transform(ax1_label_pos)
-        label_fig = fig.transFigure.inverted().transform(label_disp)
-        timeline_bounds = [ax.get_position() for ax in timeline_axes]
-        timeline_center = 0.5 * (min(pos.y0 for pos in timeline_bounds) + max(pos.y1 for pos in timeline_bounds))
-        fig.text(label_fig[0], timeline_center, 'Timeline', rotation='vertical',
-                 ha='center', va='center', fontsize=12, fontweight='bold')
+            ax2.set_xlim(-2, latent_xlim_end)
+            ax2.set_ylim(-0.4, 0.4)
+            ax2.set_xlabel('')
+            ax2.set_yticks([0])
+            ax2.set_yticklabels(['Latent Space'], fontsize=11, fontweight='bold')
+            ax2.tick_params(axis='y', length=0)
+            ax2.set_title('Latent Feature Timeline', fontsize=13, fontweight='bold')
+            ax2.grid(True, alpha=0.3, axis='x', linestyle='--')
 
-        plt.savefig(output_path, dpi=150, bbox_inches='tight')
-        # Save PDF copy for high-res needs
-        pdf_path = Path(output_path).with_suffix('.pdf')
-        plt.savefig(str(pdf_path), format='pdf', bbox_inches='tight')
-        plt.close()
+            timeline_axes = [ax2]
 
-        print(f"  ✓ {output_path.name} (PDF copy saved)")
+            if feature_mapping:
+                ax3 = fig.add_subplot(gs[2])
+                ax3.barh(0, original_frame_count, height=0.6, color='lightgray',
+                        edgecolor='black', linewidth=2)
+
+                for i, word_info in enumerate(self.word_frame_ranges):
+                    feat_start = word_info['start_frame']
+                    feat_end = word_info['end_frame']
+                    confidence = word_info['confidence']
+                    alpha = 0.9 if confidence == 'high' else 0.7 if confidence == 'medium' else 0.5
+                    if mapping_list and feat_start < len(mapping_list) and feat_end < len(mapping_list):
+                        orig_start = mapping_list[feat_start]['frame_start']
+                        orig_end = mapping_list[feat_end]['frame_end']
+                        ax3.barh(0, orig_end - orig_start, left=orig_start, height=0.6,
+                                color=colors[i % 20], alpha=alpha, edgecolor='black', linewidth=0.5)
+
+                ax3_xlim = orig_xlim_end if orig_xlim_end is not None else original_frame_count + 2
+                ax3.set_xlim(-2, ax3_xlim)
+                ax3.set_ylim(-0.4, 0.4)
+                ax3.set_xlabel('')
+                ax3.set_yticks([0])
+                ax3.set_yticklabels(['Pixel Space'], fontsize=11, fontweight='bold')
+                ax3.tick_params(axis='y', length=0)
+                ax3.set_title(f'Original Video Timeline ({original_frame_count} frames, '
+                             f'{feature_mapping["downsampling_ratio"]:.2f}x downsampling)',
+                             fontsize=13, fontweight='bold')
+                ax3.grid(True, alpha=0.3, axis='x', linestyle='--')
+                timeline_axes.append(ax3)
+                legend_row = 3
+            else:
+                legend_row = 2
+
+            ax_legend = fig.add_subplot(gs[legend_row])
+            ax_legend.axis('off')
+            legend_text = "Confidence:  ■ High (avg attn > 0.5)  ■ Medium (0.2-0.5)  ■ Low (< 0.2)"
+            ax_legend.text(0.5, 0.5, legend_text, ha='center', va='center',
+                    fontsize=11, transform=ax_legend.transAxes)
+
+            plt.tight_layout()
+            fig.canvas.draw()
+
+            ax1_pos = ax1.get_position()
+            renderer = fig.canvas.get_renderer()
+            ytick_extents = [tick.get_window_extent(renderer) for tick in ax1.get_yticklabels() if tick.get_text()]
+            fig_width_px = fig.get_size_inches()[0] * fig.dpi
+            if ytick_extents:
+                min_x_px = min(ext.x0 for ext in ytick_extents)
+            else:
+                min_x_px = ax1_pos.x0 * fig_width_px
+            line_x = max(0.01, (min_x_px / fig_width_px) - 0.01)
+            gw_center = 0.5 * (ax1_pos.y0 + ax1_pos.y1)
+            timeline_bounds = [ax.get_position() for ax in timeline_axes]
+            timeline_center = 0.5 * (min(pos.y0 for pos in timeline_bounds) + max(pos.y1 for pos in timeline_bounds))
+            fig.text(line_x, gw_center, 'Generated Word', rotation='vertical',
+                     ha='right', va='center', fontsize=12, fontweight='bold')
+            fig.text(line_x, timeline_center, 'Timeline', rotation='vertical',
+                     ha='right', va='center', fontsize=12, fontweight='bold')
+
+            png_path = Path(out_path)
+            plt.savefig(str(png_path), dpi=150, bbox_inches='tight')
+            pdf_path = png_path.with_suffix('.pdf')
+            plt.savefig(str(pdf_path), format='pdf', bbox_inches='tight')
+            plt.close()
+
+            print(f"  ✓ {png_path.name} (PDF copy saved)")
+
+        render_alignment(output_path, latent_full_limit, orig_full_limit)
+
+        if latent_short_limit < latent_full_limit - 1e-6:
+            short_path = output_path.with_name("frame_alignment_short.png")
+            render_alignment(short_path, latent_short_limit, orig_short_limit if orig_short_limit else orig_full_limit)
 
     def save_alignment_data(self, output_path):
         """保存帧对齐数据为JSON"""