{ "table_of_contents": [ { "title": "Diffusion Model-Based Image Editing: A Survey", "heading_level": null, "page_id": 0, "polygon": [ [ 51.70507192611694, 52.462727308273315 ], [ 559.7051510810852, 52.462727308273315 ], [ 559.7051510810852, 79.77545166015625 ], [ 51.70507192611694, 79.77545166015625 ] ] }, { "title": "1 INTRODUCTION", "heading_level": null, "page_id": 0, "polygon": [ [ 48.275001525878906, 299.7940673828125 ], [ 139.82582092285156, 299.24456906318665 ], [ 139.82582092285156, 310.7940673828125 ], [ 48.275001525878906, 311.097980260849 ] ] }, { "title": "2 BACKGROUND", "heading_level": null, "page_id": 1, "polygon": [ [ 47.96636438369751, 637.3428192138672 ], [ 137.5156545639038, 636.9589676856995 ], [ 137.5156545639038, 648.3428192138672 ], [ 47.96636438369751, 648.7333073616028 ] ] }, { "title": "2.1 Diffusion Models", "heading_level": null, "page_id": 1, "polygon": [ [ 48.0, 653.0731658935547 ], [ 149.0821123123169, 651.7939653396606 ], [ 149.0821123123169, 662.5731658935547 ], [ 48.0, 663.1411771774292 ] ] }, { "title": "2.2 Related Tasks", "heading_level": null, "page_id": 2, "polygon": [ [ 311.06041646003723, 123.75128173828125 ], [ 398.50966572761536, 123.53123152256012 ], [ 397.8325500488281, 133.25128173828125 ], [ 309.86510396003723, 134.0058478116989 ] ] }, { "title": "2.2.1 Conditional Image Generation", "heading_level": null, "page_id": 2, "polygon": [ [ 312.00006103515625, 141.0194091796875 ], [ 469.5709354877472, 141.0194091796875 ], [ 468.7310485839844, 150.5194091796875 ], [ 310.9420816898346, 151.90257740020752 ] ] }, { "title": "2.2.2 Image Restoration and Enhancement", "heading_level": null, "page_id": 3, "polygon": [ [ 47.53681719303131, 93.17786353826523 ], [ 235.89097595214844, 93.17786353826523 ], [ 235.89097595214844, 103.7764892578125 ], [ 47.53681719303131, 103.7764892578125 ] ] }, { "title": "3 CATEGORIZATION OF IMAGE EDITING", "heading_level": null, "page_id": 3, "polygon": [ [ 312.2749938964844, 217.53857421875 ], [ 511.4214172363281, 217.53857421875 ], [ 511.4214172363281, 228.53857421875 ], [ 312.2749938964844, 228.53857421875 ] ] }, { "title": "TABLE 1 (continued)", "heading_level": null, "page_id": 5, "polygon": [ [ 47.45294961333275, 38.46849596500397 ], [ 132.04697585105896, 38.46849596500397 ], [ 132.04697585105896, 49.01025390625 ], [ 47.45294961333275, 49.01025390625 ] ] }, { "title": "4 TRAINING-BASED APPROACHES", "heading_level": null, "page_id": 6, "polygon": [ [ 46.88917690515518, 231.01441383361816 ], [ 224.04689741134644, 231.01441383361816 ], [ 224.04689741134644, 242.69537830352783 ], [ 46.88917690515518, 242.69537830352783 ] ] }, { "title": "4.1 Domain-Specific Editing with Weak Supervision", "heading_level": null, "page_id": 6, "polygon": [ [ 47.06102496385574, 399.24906063079834 ], [ 285.8515319824219, 397.70218563079834 ], [ 285.8515319824219, 408.89044189453125 ], [ 47.06102496385574, 409.452449798584 ] ] }, { "title": "4.2 Reference and Attribute Guidance via Self-\nSupervision", "heading_level": null, "page_id": 7, "polygon": [ [ 48.00000762939453, 455.1441650390625 ], [ 300.18102264404297, 455.1441650390625 ], [ 299.9970397949219, 476.40842056274414 ], [ 47.48034006357193, 476.40842056274414 ] ] }, { "title": "4.3 Instructional Editing via Full Supervision", "heading_level": null, "page_id": 7, "polygon": [ [ 309.6724998950958, 331.0684254169464 ], [ 519.6997182369232, 329.5215504169464 ], [ 519.6997182369232, 340.7367858886719 ], [ 309.6724998950958, 340.74592995643616 ] ] }, { "title": "4.4 Pseudo-Target Retrieval with Weak Supervision", "heading_level": null, "page_id": 8, "polygon": [ [ 311.1615700721741, 628.9691467285156 ], [ 548.7210083007812, 628.9691467285156 ], [ 548.7210083007812, 638.4691467285156 ], [ 311.1615700721741, 638.4691467285156 ] ] }, { "title": "5 TESTING-TIME FINETUNING APPROACHES", "heading_level": null, "page_id": 9, "polygon": [ [ 47.880093812942505, 357.18902587890625 ], [ 272.29241943359375, 357.18902587890625 ], [ 272.29241943359375, 368.18902587890625 ], [ 47.880093812942505, 368.18902587890625 ] ] }, { "title": "5.1 Denoising Model Finetuning", "heading_level": null, "page_id": 9, "polygon": [ [ 47.512673288583755, 525.4051008224487 ], [ 197.91000366210938, 525.4051008224487 ], [ 197.91000366210938, 535.6764831542969 ], [ 47.512673288583755, 535.6764831542969 ] ] }, { "title": "5.2 Embedding Finetuning", "heading_level": null, "page_id": 9, "polygon": [ [ 311.9808669090271, 220.4031982421875 ], [ 438.1999282836914, 220.4031982421875 ], [ 438.1999282836914, 230.71972489356995 ], [ 311.9808669090271, 230.71972489356995 ] ] }, { "title": "5.3 Guidance with a Hypernetwork", "heading_level": null, "page_id": 9, "polygon": [ [ 311.45996046066284, 664.2454876899719 ], [ 475.2201805114746, 664.2454876899719 ], [ 474.0248680114746, 674.1751861572266 ], [ 310.26464796066284, 674.1751861572266 ] ] }, { "title": "5.4 Latent Variable Optimization", "heading_level": null, "page_id": 10, "polygon": [ [ 47.53924298286438, 330.6864318847656 ], [ 198.31851196289062, 329.16808891296387 ], [ 198.31851196289062, 340.1864318847656 ], [ 47.53924298286438, 340.1864318847656 ] ] }, { "title": "5.5 Hybrid Finetuning", "heading_level": null, "page_id": 10, "polygon": [ [ 311.0644655227661, 299.1764221191406 ], [ 415.98699951171875, 299.1764221191406 ], [ 415.98699951171875, 309.46584963798523 ], [ 311.0644655227661, 309.46584963798523 ] ] }, { "title": "6 TRAINING AND FINETUNING FREE APPROACHES", "heading_level": null, "page_id": 10, "polygon": [ [ 311.3416256904602, 674.920331954956 ], [ 563.7304077148438, 674.920331954956 ], [ 563.7304077148438, 686.0591125488281 ], [ 311.3416256904602, 686.0591125488281 ] ] }, { "title": "6.1 Input Text Refinement", "heading_level": null, "page_id": 11, "polygon": [ [ 47.15997165441513, 318.4724426269531 ], [ 169.84402799606323, 317.8506646156311 ], [ 169.84402799606323, 327.97781467437744 ], [ 47.15997165441513, 329.52468967437744 ] ] }, { "title": "6.2 Inversion/Sampling Modification", "heading_level": null, "page_id": 11, "polygon": [ [ 310.1737446784973, 268.731458902359 ], [ 483.41526460647583, 268.731458902359 ], [ 482.21995210647583, 278.57753920555115 ], [ 308.9784321784973, 278.57753920555115 ] ] }, { "title": "6.3 Attention Modification", "heading_level": null, "page_id": 12, "polygon": [ [ 47.614460706710815, 490.04156494140625 ], [ 170.97747802734375, 488.81344413757324 ], [ 170.97747802734375, 499.54156494140625 ], [ 47.614460706710815, 499.65016078948975 ] ] }, { "title": "6.4 Mask Guidance", "heading_level": null, "page_id": 12, "polygon": [ [ 310.74320340156555, 686.0300931930542 ], [ 405.57918334007263, 686.0300931930542 ], [ 405.57918334007263, 696.1405487060547 ], [ 310.74320340156555, 696.1405487060547 ] ] }, { "title": "6.5 Multi-Noise Redirection", "heading_level": null, "page_id": 13, "polygon": [ [ 47.643985122442245, 642.1881713867188 ], [ 176.80099487304688, 641.1558480262756 ], [ 176.80099487304688, 651.6881713867188 ], [ 47.643985122442245, 652.4922022819519 ] ] }, { "title": "7 INPAINTING AND OUTPAINTING", "heading_level": null, "page_id": 13, "polygon": [ [ 310.7379870414734, 442.0059814453125 ], [ 479.63800048828125, 441.63871335983276 ], [ 479.63800048828125, 453.0059814453125 ], [ 310.7379870414734, 453.64493322372437 ] ] }, { "title": "7.1 Traditional Context-Driven Inpainting", "heading_level": null, "page_id": 13, "polygon": [ [ 311.10400772094727, 520.4220843315125 ], [ 501.8072018623352, 518.8752093315125 ], [ 501.12603759765625, 530.1444091796875 ], [ 309.90869522094727, 530.3065438270569 ] ] }, { "title": "7.1.1 Supervised Training", "heading_level": null, "page_id": 13, "polygon": [ [ 310.78062987327576, 535.6135559082031 ], [ 428.4993588924408, 534.4718899726868 ], [ 428.4993588924408, 545.1135406494141 ], [ 310.78062987327576, 545.3104948997498 ] ] }, { "title": "7.1.2 Zero-Shot Learning", "heading_level": null, "page_id": 13, "polygon": [ [ 310.36372232437134, 711.2014846801758 ], [ 425.13400411605835, 711.2014846801758 ], [ 425.13400411605835, 721.265118598938 ], [ 310.36372232437134, 721.265118598938 ] ] }, { "title": "7.2 Multimodal Conditional Inpainting", "heading_level": null, "page_id": 14, "polygon": [ [ 310.59240317344666, 140.727294921875 ], [ 488.19261145591736, 140.727294921875 ], [ 487.7689514160156, 150.3015797138214 ], [ 309.39709067344666, 150.3015797138214 ] ] }, { "title": "7.2.1 Random Mask Training", "heading_level": null, "page_id": 14, "polygon": [ [ 311.8589208126068, 155.46441650390625 ], [ 440.56935954093933, 155.46441650390625 ], [ 440.56935954093933, 165.16435861587524 ], [ 311.8589208126068, 165.16435861587524 ] ] }, { "title": "7.2.2 Precise Control Conditioning", "heading_level": null, "page_id": 14, "polygon": [ [ 310.11118483543396, 399.66534423828125 ], [ 464.56235432624817, 399.66534423828125 ], [ 464.56235432624817, 409.59364557266235 ], [ 310.11118483543396, 409.59364557266235 ] ] }, { "title": "7.2.3 Pretrained Diffusion Prior Exploiting", "heading_level": null, "page_id": 15, "polygon": [ [ 48.0, 183.0263671875 ], [ 228.61398315429688, 182.38901567459106 ], [ 228.61398315429688, 192.5263671875 ], [ 48.0, 193.5333559513092 ] ] }, { "title": "7.3 Outpainting", "heading_level": null, "page_id": 15, "polygon": [ [ 47.494689613580704, 606.3902740478516 ], [ 124.55635571479797, 605.377046585083 ], [ 124.55635571479797, 615.8902740478516 ], [ 47.494689613580704, 616.374077796936 ] ] }, { "title": "8 BENCHMARK AND EVALUATION", "heading_level": null, "page_id": 15, "polygon": [ [ 312.2749938964844, 93.99383544921875 ], [ 483.1845774650574, 93.74228078126907 ], [ 482.53759765625, 104.99383544921875 ], [ 311.33053636550903, 106.22412765026093 ] ] }, { "title": "8.1 Benchmark Construction", "heading_level": null, "page_id": 15, "polygon": [ [ 312.0, 109.62322998046875 ], [ 448.09698486328125, 108.45888948440552 ], [ 448.09698486328125, 119.12322998046875 ], [ 312.0, 119.65671515464783 ] ] }, { "title": "8.1.1 Task Selection", "heading_level": null, "page_id": 15, "polygon": [ [ 311.3871502876282, 445.70623779296875 ], [ 404.0688829421997, 444.72826194763184 ], [ 404.0688829421997, 455.20623779296875 ], [ 311.3871502876282, 455.6185111999512 ] ] }, { "title": "8.1.2 Dataset Construction", "heading_level": null, "page_id": 15, "polygon": [ [ 311.8153660297394, 630.9782409667969 ], [ 431.4720153808594, 630.0558114051819 ], [ 431.4720153808594, 640.4782409667969 ], [ 311.8153660297394, 641.0521817207336 ] ] }, { "title": "8.1.3 Metric Design and Selection", "heading_level": null, "page_id": 16, "polygon": [ [ 48.0, 285.8451843261719 ], [ 197.72558212280273, 285.8451843261719 ], [ 197.72558212280273, 295.3451843261719 ], [ 48.0, 295.3451843261719 ] ] }, { "title": "8.2 Evaluation", "heading_level": null, "page_id": 16, "polygon": [ [ 310.56953144073486, 337.0557556152344 ], [ 383.1181569099426, 335.6536295413971 ], [ 383.1181569099426, 346.5557556152344 ], [ 310.56953144073486, 346.5557556152344 ] ] }, { "title": "8.2.1 Method Selection", "heading_level": null, "page_id": 16, "polygon": [ [ 311.5959506034851, 351.8018798828125 ], [ 416.8720192909241, 351.8018798828125 ], [ 416.8720192909241, 361.3018798828125 ], [ 311.5959506034851, 361.3018798828125 ] ] }, { "title": "8.2.2 Comparative Analysis", "heading_level": null, "page_id": 16, "polygon": [ [ 312.0000915527344, 492.1639709472656 ], [ 434.1510925292969, 492.1639709472656 ], [ 434.1510925292969, 501.6639709472656 ], [ 312.0000915527344, 501.6639709472656 ] ] }, { "title": "Editing Type: Object Addition (\"Add a glass of milk next to the cookies\")", "heading_level": null, "page_id": 18, "polygon": [ [ 152.6357211470604, 74.15595799684525 ], [ 453.2707586288452, 74.15595799684525 ], [ 453.2707586288452, 84.79154974222183 ], [ 152.6357211470604, 84.79154974222183 ] ] }, { "title": "9 CHALLENGES AND FUTURE DIRECTIONS", "heading_level": null, "page_id": 19, "polygon": [ [ 47.049424946308136, 269.604736328125 ], [ 264.90347242355347, 269.604736328125 ], [ 264.90347242355347, 280.604736328125 ], [ 47.049424946308136, 281.4705419540405 ] ] }, { "title": "10 CONCLUSION", "heading_level": null, "page_id": 19, "polygon": [ [ 311.0942680835724, 558.6276397705078 ], [ 402.31228280067444, 558.6276397705078 ], [ 402.31228280067444, 569.6276397705078 ], [ 311.0942680835724, 569.6276397705078 ] ] }, { "title": "REFERENCES", "heading_level": null, "page_id": 20, "polygon": [ [ 48.1486998796463, 43.31805419921875 ], [ 115.23244035243988, 42.40864169597626 ], [ 115.23244035243988, 54.31805419921875 ], [ 48.1486998796463, 54.529072701931 ] ] } ], "page_stats": [ { "page_id": 0, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 157 ], [ "Line", 68 ], [ "Text", 7 ], [ "ListItem", 6 ], [ "SectionHeader", 2 ], [ "Figure", 1 ], [ "Caption", 1 ], [ "PageHeader", 1 ], [ "FigureGroup", 1 ], [ "ListGroup", 1 ] ] }, { "page_id": 1, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 784 ], [ "Line", 131 ], [ "Text", 6 ], [ "TextInlineMath", 6 ], [ "Equation", 6 ], [ "SectionHeader", 2 ], [ "PageHeader", 1 ] ] }, { "page_id": 2, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 811 ], [ "Line", 149 ], [ "Text", 10 ], [ "Equation", 6 ], [ "TextInlineMath", 5 ], [ "SectionHeader", 2 ], [ "PageHeader", 1 ] ] }, { "page_id": 3, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 266 ], [ "Line", 119 ], [ "Text", 11 ], [ "ListItem", 3 ], [ "SectionHeader", 2 ], [ "PageHeader", 1 ], [ "ListGroup", 1 ] ] }, { "page_id": 4, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 643 ], [ "Line", 92 ], [ "Text", 2 ], [ "Table", 1 ], [ "PageHeader", 1 ] ] }, { "page_id": 5, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 813 ], [ "Line", 93 ], [ "SectionHeader", 1 ], [ "Table", 1 ], [ "PageHeader", 1 ] ] }, { "page_id": 6, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 176 ], [ "Line", 85 ], [ "Text", 7 ], [ "Figure", 3 ], [ "Caption", 2 ], [ "SectionHeader", 2 ], [ "FigureGroup", 2 ], [ "PageHeader", 1 ] ] }, { "page_id": 7, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 244 ], [ "Line", 120 ], [ "Text", 12 ], [ "SectionHeader", 2 ], [ "PageHeader", 1 ] ] }, { "page_id": 8, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 186 ], [ "Line", 95 ], [ "Text", 11 ], [ "Figure", 1 ], [ "Caption", 1 ], [ "SectionHeader", 1 ], [ "PageHeader", 1 ], [ "FigureGroup", 1 ] ] }, { "page_id": 9, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 212 ], [ "Line", 103 ], [ "Text", 10 ], [ "SectionHeader", 4 ], [ "Figure", 1 ], [ "Caption", 1 ], [ "PageHeader", 1 ], [ "FigureGroup", 1 ] ] }, { "page_id": 10, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 220 ], [ "Line", 104 ], [ "Text", 9 ], [ "SectionHeader", 3 ], [ "PageHeader", 1 ], [ "Figure", 1 ], [ "Caption", 1 ], [ "FigureGroup", 1 ] ] }, { "page_id": 11, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 211 ], [ "Line", 106 ], [ "Text", 7 ], [ "SectionHeader", 2 ], [ "PageHeader", 1 ], [ "Figure", 1 ], [ "Caption", 1 ], [ "FigureGroup", 1 ] ] }, { "page_id": 12, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 164 ], [ "Line", 85 ], [ "Text", 9 ], [ "PageHeader", 2 ], [ "SectionHeader", 2 ], [ "Figure", 1 ], [ "Caption", 1 ], [ "FigureGroup", 1 ] ] }, { "page_id": 13, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 222 ], [ "Line", 105 ], [ "Text", 11 ], [ "SectionHeader", 5 ], [ "Picture", 1 ], [ "Caption", 1 ], [ "PageHeader", 1 ], [ "PictureGroup", 1 ] ] }, { "page_id": 14, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 619 ], [ "Line", 134 ], [ "Text", 8 ], [ "Equation", 4 ], [ "TextInlineMath", 4 ], [ "SectionHeader", 3 ], [ "PageHeader", 1 ] ] }, { "page_id": 15, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 218 ], [ "Line", 116 ], [ "Text", 12 ], [ "SectionHeader", 6 ], [ "PageHeader", 1 ] ] }, { "page_id": 16, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 399 ], [ "Line", 129 ], [ "Text", 6 ], [ "SectionHeader", 4 ], [ "TextInlineMath", 4 ], [ "ListItem", 4 ], [ "Equation", 1 ], [ "PageHeader", 1 ], [ "ListGroup", 1 ] ] }, { "page_id": 17, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 85 ], [ "Line", 32 ], [ "Figure", 3 ], [ "Caption", 3 ], [ "Text", 3 ], [ "FigureGroup", 3 ], [ "PageHeader", 1 ] ] }, { "page_id": 18, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 185 ], [ "Line", 36 ], [ "Text", 24 ], [ "Picture", 11 ], [ "Caption", 4 ], [ "PictureGroup", 3 ], [ "PageHeader", 1 ], [ "SectionHeader", 1 ], [ "Table", 1 ] ] }, { "page_id": 19, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 283 ], [ "Line", 121 ], [ "Text", 8 ], [ "SectionHeader", 2 ], [ "TextInlineMath", 1 ], [ "PageHeader", 1 ] ] }, { "page_id": 20, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 583 ], [ "Line", 153 ], [ "ListItem", 55 ], [ "ListGroup", 2 ], [ "SectionHeader", 1 ], [ "PageHeader", 1 ] ] }, { "page_id": 21, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 573 ], [ "Line", 153 ], [ "ListItem", 50 ], [ "ListGroup", 2 ], [ "Text", 1 ], [ "PageHeader", 1 ] ] }, { "page_id": 22, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 570 ], [ "Line", 154 ], [ "ListItem", 53 ], [ "ListGroup", 2 ], [ "PageHeader", 1 ] ] }, { "page_id": 23, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 558 ], [ "Line", 154 ], [ "ListItem", 54 ], [ "ListGroup", 2 ], [ "PageHeader", 1 ] ] }, { "page_id": 24, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 578 ], [ "Line", 154 ], [ "ListItem", 54 ], [ "ListGroup", 2 ], [ "PageHeader", 1 ] ] }, { "page_id": 25, "text_extraction_method": "pdftext", "block_counts": [ [ "Span", 326 ], [ "Line", 89 ], [ "ListItem", 31 ], [ "ListGroup", 2 ], [ "PageHeader", 1 ] ] } ], "debug_data_path": "debug_data\\Huang \u7b49 - 2024 - Diffusion model-based image editing a survey" }