{"noiseShare":0.4824,"topics":[{"claims":[224,null,320,null,389,null,413],"deltaPp":-3.4971,"id":0,"keywords":["segmentation","semantic","instance","state-of-the-art","classification"],"label":"Generalist Segmentation","labelFull":"Generalist Segmentation Models for Multiple Segmentation Tasks","papers":677,"papersByYear":[116,null,165,null,199,null,197],"rel":-0.3241,"reviewed":true,"share":[10.7907,null,10.2357,null,9.2301,null,7.2936],"size":1346,"slope":-0.005748},{"claims":[124,null,212,null,210,null,227],"deltaPp":-1.0789,"id":1,"keywords":["adversarial","attack","backdoor","examples","robustness"],"label":"Adversarial Training","labelFull":"Adversarial Training Methods for Improving Model Robustness","papers":307,"papersByYear":[51,null,79,null,78,null,99],"rel":-0.2274,"reviewed":true,"share":[4.7442,null,4.9007,null,3.6178,null,3.6653],"size":773,"slope":-0.00226},{"claims":[133,null,148,null,173,null,191],"deltaPp":-1.8781,"id":2,"keywords":["depth","stereo","monocular","estimation","disparity"],"label":"Monocular Depth Estimation","labelFull":"Monocular Depth Estimation and Self-Supervised Learning","papers":296,"papersByYear":[58,null,66,null,77,null,95],"rel":-0.3481,"reviewed":true,"share":[5.3953,null,4.0943,null,3.5714,null,3.5172],"size":645,"slope":-0.003079},{"claims":[134,null,126,null,170,null,176],"deltaPp":-2.1896,"id":3,"keywords":["face","facial","identity","expression","recognition"],"label":"Disentangled Face Representation","labelFull":"Disentangled Face Representation for Recognition and Editing","papers":230,"papersByYear":[51,null,42,null,68,null,69],"rel":-0.4615,"reviewed":true,"share":[4.7442,null,2.6055,null,3.154,null,2.5546],"size":606,"slope":-0.00301},{"claims":[113,null,136,null,173,null,102],"deltaPp":-3.2128,"id":4,"keywords":["object detection","detectors","coco","localization","domain"],"label":"Generalized Object Detection","labelFull":"Object Detection Beyond Fully Supervised Settings","papers":307,"papersByYear":[62,null,72,null,104,null,69],"rel":-0.5571,"reviewed":true,"share":[5.7674,null,4.4665,null,4.8237,null,2.5546],"size":524,"slope":-0.004641},{"claims":[76,null,148,null,136,null,147],"deltaPp":-0.3726,"id":5,"keywords":["video","temporal","frames","grounding","features"],"label":"Temporal Video Modeling","labelFull":"Temporal Video Representation Learning for Vision Tasks","papers":333,"papersByYear":[45,null,91,null,94,null,103],"rel":-0.089,"reviewed":true,"share":[4.186,null,5.6452,null,4.3599,null,3.8134],"size":507,"slope":-0.001202},{"claims":[104,null,125,null,116,null,130],"deltaPp":-1.8356,"id":6,"keywords":["action","recognition","egocentric","temporal","skeleton-based"],"label":"Video Action Recognition","labelFull":"Video Action Recognition, Localization, and Segmentation","papers":241,"papersByYear":[46,null,63,null,66,null,66],"rel":-0.429,"reviewed":true,"share":[4.2791,null,3.9082,null,3.0612,null,2.4435],"size":475,"slope":-0.003177},{"claims":[98,null,143,null,124,null,84],"deltaPp":-3.0222,"id":7,"keywords":["pose","estimation","human","3d","6d"],"label":"3D/6D Pose Estimation","labelFull":"3D and 6D Pose Estimation for Human and Object Understanding","papers":234,"papersByYear":[50,null,73,null,67,null,44],"rel":-0.6498,"reviewed":true,"share":[4.6512,null,4.5285,null,3.1076,null,1.629],"size":449,"slope":-0.005244},{"claims":[51,null,126,null,117,null,107],"deltaPp":0.0439,"id":8,"keywords":["point","cloud","registration","lidar","completion"],"label":"3D Point Cloud Deep Learning","labelFull":"Deep Learning for 3D Point Cloud Analysis","papers":219,"papersByYear":[25,null,59,null,71,null,64],"rel":0.0189,"reviewed":true,"share":[2.3256,null,3.66,null,3.2931,null,2.3695],"size":401,"slope":-0.000118},{"claims":[56,null,87,null,102,null,130],"deltaPp":-0.3584,"id":9,"keywords":["state-of-the-art","benchmarks","outperforms","superior","compared"],"label":"SOTA Performance Claims","labelFull":"State-of-the-Art Performance Claims on Benchmarks","papers":361,"papersByYear":[54,null,82,null,99,null,126],"rel":-0.0713,"reviewed":true,"share":[5.0233,null,5.0868,null,4.5918,null,4.6649],"size":375,"slope":-0.000785},{"claims":[35,null,56,null,140,null,128],"deltaPp":0.5678,"id":10,"keywords":["audio","speech","sound","lip","talking"],"label":"Audio-Driven Visual Generation","labelFull":"Audio-Driven Generation of Talking Heads, Gestures, Lip Sync","papers":124,"papersByYear":[13,null,21,null,42,null,48],"rel":0.4695,"reviewed":true,"share":[1.2093,null,1.3027,null,1.9481,null,1.7771],"size":359,"slope":0.001174},{"claims":[65,null,112,null,113,null,49],"deltaPp":-1.997,"id":11,"keywords":["domain","adaptation","target","source","generalization"],"label":"Unsupervised Domain Adaptation","labelFull":"Unsupervised, Multi-Source, and Source-Free Domain Adaptation","papers":193,"papersByYear":[35,null,56,null,68,null,34],"rel":-0.6134,"reviewed":true,"share":[3.2558,null,3.4739,null,3.154,null,1.2588],"size":339,"slope":-0.003155},{"claims":[27,null,56,null,94,null,150],"deltaPp":1.7326,"id":12,"keywords":["motion","human","prediction","generation","sequences"],"label":"Human Motion Synthesis","labelFull":"Human Motion Generation and Prediction","papers":192,"papersByYear":[16,null,28,null,61,null,87],"rel":1.1641,"reviewed":true,"share":[1.4884,null,1.737,null,2.8293,null,3.221],"size":327,"slope":0.003145},{"claims":[85,null,74,null,80,null,59],"deltaPp":-3.4331,"id":13,"keywords":["architecture search","neural architecture","networks","convolutional","parameters"],"label":"Training-Efficient Neural Architecture Search","labelFull":"Neural Architecture Search for Training-Efficient Networks","papers":231,"papersByYear":[58,null,57,null,63,null,53],"rel":-0.6363,"reviewed":true,"share":[5.3953,null,3.536,null,2.9221,null,1.9622],"size":298,"slope":-0.005457},{"claims":[10,null,7,null,71,null,205],"deltaPp":3.9027,"id":14,"keywords":["vision-language","visual","decoding","tokens","pre-trained"],"label":"Efficient Vision-Language Models","labelFull":"Efficient Training and Inference For Vision-Language Models","papers":187,"papersByYear":[7,null,5,null,52,null,123],"rel":5.9931,"reviewed":true,"share":[0.6512,null,0.3102,null,2.4119,null,4.5539],"size":293,"slope":0.006905},{"claims":[53,null,64,null,93,null,77],"deltaPp":-0.8067,"id":15,"keywords":["tracking","trackers","multi-object","association","target"],"label":"Multi-Object Tracking","labelFull":"Multi-Object Tracking and Association Methods","papers":125,"papersByYear":[23,null,28,null,38,null,36],"rel":-0.3771,"reviewed":true,"share":[2.1395,null,1.737,null,1.7625,null,1.3328],"size":287,"slope":-0.001197},{"claims":[32,null,40,null,90,null,113],"deltaPp":1.1783,"id":16,"keywords":["distillation","knowledge","student","teacher","loss"],"label":"Knowledge Distillation","labelFull":"Knowledge Distillation for Computer Vision","papers":155,"papersByYear":[14,null,25,null,49,null,67],"rel":0.9048,"reviewed":true,"share":[1.3023,null,1.5509,null,2.2727,null,2.4806],"size":275,"slope":0.002128},{"claims":[1,null,3,null,63,null,180],"deltaPp":4.3128,"id":17,"keywords":["diffusion","latent","generative","process","inference"],"label":"Efficient Diffusion Models","labelFull":"Efficient Diffusion Model Training and Inference","papers":161,"papersByYear":[1,null,2,null,39,null,119],"rel":46.3742,"reviewed":true,"share":[0.093,null,0.1241,null,1.8089,null,4.4058],"size":247,"slope":0.007312},{"claims":[35,null,38,null,39,null,134],"deltaPp":0.9931,"id":18,"keywords":["driving","autonomous","vehicle","traffic","driver"],"label":"End-to-End Driving Benchmarks","labelFull":"Datasets and Evaluation Platforms for End-to-End Autonomous Driving","papers":128,"papersByYear":[14,null,26,null,26,null,62],"rel":0.7626,"reviewed":true,"share":[1.3023,null,1.6129,null,1.2059,null,2.2954],"size":246,"slope":0.001286},{"claims":[11,null,28,null,89,null,105],"deltaPp":1.238,"id":19,"keywords":["anomaly","detection","out-of-distribution","ood","in-distribution"],"label":"Visual Anomaly Detection","labelFull":"Visual Anomaly Detection and Localization","papers":103,"papersByYear":[5,null,12,null,40,null,46],"rel":2.6618,"reviewed":true,"share":[0.4651,null,0.7444,null,1.8553,null,1.7031],"size":233,"slope":0.002412},{"claims":[14,null,32,null,47,null,138],"deltaPp":2.2908,"id":20,"keywords":["modality","multimodal","fusion","shared","representations"],"label":"Shared Multimodal Representations","labelFull":"Learning Shared Representations for Multimodal Fusion","papers":152,"papersByYear":[10,null,24,null,31,null,87],"rel":2.4627,"reviewed":true,"share":[0.9302,null,1.4888,null,1.4378,null,3.221],"size":231,"slope":0.003411},{"claims":[77,null,75,null,27,null,51],"deltaPp":-2.5343,"id":21,"keywords":["person","re-identification","visible-infrared","search","unsupervised"],"label":"Person Re-Identification","labelFull":"Deep Learning Approaches for Person Re-Identification","papers":105,"papersByYear":[36,null,34,null,13,null,22],"rel":-0.7568,"reviewed":true,"share":[3.3488,null,2.1092,null,0.603,null,0.8145],"size":230,"slope":-0.004555},{"claims":[49,null,62,null,45,null,73],"deltaPp":-0.4564,"id":22,"keywords":["few-shot","zero-shot","learning","classification","compositional"],"label":"Few-shot & Zero-shot Learning","labelFull":"Few-shot and Zero-shot Classification in Vision","papers":156,"papersByYear":[26,null,40,null,37,null,53],"rel":-0.1887,"reviewed":true,"share":[2.4186,null,2.4814,null,1.7161,null,1.9622],"size":229,"slope":-0.001067},{"claims":[29,null,52,null,57,null,91],"deltaPp":1.0861,"id":23,"keywords":["question","answering","visual","vqa","reasoning"],"label":"Visual Question Answering","labelFull":"Visual Question Answering: Datasets, Benchmarks, and Advances","papers":122,"papersByYear":[13,null,23,null,24,null,62],"rel":0.8981,"reviewed":true,"share":[1.2093,null,1.4268,null,1.1132,null,2.2954],"size":229,"slope":0.001472},{"claims":[34,null,62,null,89,null,38],"deltaPp":-1.1779,"id":24,"keywords":["label","semi-supervised learning","unlabeled","noisy","pseudo"],"label":"Learning With Noisy Labels","labelFull":"Noisy Label Learning And Semi-supervised Pseudo-Labeling","papers":159,"papersByYear":[25,null,44,null,59,null,31],"rel":-0.5065,"reviewed":true,"share":[2.3256,null,2.7295,null,2.7365,null,1.1477],"size":223,"slope":-0.001763},{"claims":[7,null,19,null,55,null,130],"deltaPp":1.6461,"id":25,"keywords":["affordance","manipulation","robotic","grasp","dexterous"],"label":"Affordance-Based Dexterous Manipulation","labelFull":"Affordance-Based Dexterous Manipulation and Grasp Synthesis","papers":83,"papersByYear":[3,null,10,null,18,null,52],"rel":5.8979,"reviewed":true,"share":[0.2791,null,0.6203,null,0.8349,null,1.9252],"size":211,"slope":0.002576},{"claims":[38,null,45,null,45,null,77],"deltaPp":0.0106,"id":26,"keywords":["compression","jpeg","coding","codecs","psnr"],"label":"Neural Image Compression","labelFull":"Neural Image Compression with Variable Rate and RD Loss","papers":102,"papersByYear":[17,null,23,null,19,null,43],"rel":0.0067,"reviewed":true,"share":[1.5814,null,1.4268,null,0.8813,null,1.592],"size":205,"slope":-0.000257},{"claims":[0,null,3,null,59,null,142],"deltaPp":3.4061,"id":27,"keywords":["prompt","tuning","visual","instruction","text"],"label":"Vision-Language Prompt Tuning","labelFull":"Multimodal Prompt Learning for Vision-Language Models","papers":130,"papersByYear":[0,null,3,null,35,null,92],"rel":null,"reviewed":true,"share":[0.0,null,0.1861,null,1.6234,null,3.4061],"size":204,"slope":0.005828},{"claims":[33,null,47,null,46,null,75],"deltaPp":0.0314,"id":28,"keywords":["style","stylization","transfer","content","artistic"],"label":"Neural Style Transfer","labelFull":"Neural Image and Video Style Transfer","papers":85,"papersByYear":[12,null,21,null,21,null,31],"rel":0.0281,"reviewed":true,"share":[1.1163,null,1.3027,null,0.974,null,1.1477],"size":201,"slope":-0.000117},{"claims":[60,null,35,null,52,null,53],"deltaPp":-1.1959,"id":29,"keywords":["captioning","change","image","pairs","generated"],"label":"Visual Captioning","labelFull":"Image and Video Captioning with Large Language Models","papers":93,"papersByYear":[24,null,17,null,24,null,28],"rel":-0.5357,"reviewed":true,"share":[2.2326,null,1.0546,null,1.1132,null,1.0367],"size":200,"slope":-0.001765},{"claims":[33,null,50,null,68,null,46],"deltaPp":-0.6758,"id":30,"keywords":["retrieval","image","text-video","composed","benchmarks"],"label":"Cross-Modal Retrieval","labelFull":"Cross-Modal Image and Video Retrieval","papers":122,"papersByYear":[20,null,28,null,42,null,32],"rel":-0.3632,"reviewed":true,"share":[1.8605,null,1.737,null,1.9481,null,1.1847],"size":197,"slope":-0.000908},{"claims":[5,null,15,null,47,null,128],"deltaPp":3.2751,"id":31,"keywords":["editing","generation","text-to-image","instruction-based","font"],"label":"Controllable Text-to-Image Generation","labelFull":"Controllable Text-to-Image Generation and Editing","papers":136,"papersByYear":[3,null,7,null,30,null,96],"rel":11.7345,"reviewed":true,"share":[0.2791,null,0.4342,null,1.3915,null,3.5542],"size":195,"slope":0.005391},{"claims":[37,null,41,null,41,null,72],"deltaPp":-0.0254,"id":32,"keywords":["inpainting","outpainting","image","regions","video"],"label":"Image and Video Inpainting","labelFull":"Deep Learning Approaches for Image and Video Inpainting","papers":91,"papersByYear":[15,null,19,null,20,null,37],"rel":-0.0182,"reviewed":true,"share":[1.3953,null,1.1787,null,0.9276,null,1.3699],"size":191,"slope":-0.000164},{"claims":[26,null,49,null,52,null,62],"deltaPp":-0.4166,"id":33,"keywords":["flow","optical","scene","estimation","motion"],"label":"Optical Flow Estimation","labelFull":"Optical Flow Estimation, Datasets, and High-Res Methods","papers":112,"papersByYear":[20,null,27,null,26,null,39],"rel":-0.2239,"reviewed":true,"share":[1.8605,null,1.6749,null,1.2059,null,1.4439],"size":189,"slope":-0.000859},{"claims":[55,null,46,null,39,null,45],"deltaPp":-2.2011,"id":34,"keywords":["algorithm","optimization","convergence","solution","minimal"],"label":"Efficient Optimization Algorithms","labelFull":"Fast Low-Resource Optimization Algorithms for Vision","papers":127,"papersByYear":[36,null,33,null,27,null,31],"rel":-0.6573,"reviewed":true,"share":[3.3488,null,2.0471,null,1.2523,null,1.1477],"size":185,"slope":-0.003699},{"claims":[43,null,47,null,44,null,50],"deltaPp":-0.9918,"id":35,"keywords":["camera","rotation","epipolar","solvers","minimal"],"label":"Minimal Camera Geometry Solvers","labelFull":"Minimal Solvers for Camera Calibration, Pose, and Distortion","papers":102,"papersByYear":[23,null,22,null,26,null,31],"rel":-0.4636,"reviewed":true,"share":[2.1395,null,1.3648,null,1.2059,null,1.1477],"size":184,"slope":-0.001567},{"claims":[8,null,10,null,53,null,112],"deltaPp":2.3857,"id":36,"keywords":["rendering","differentiable","inverse","quality","gaussian"],"label":"Gaussian Splatting","labelFull":"Gaussian Splatting for Real-Time High-Quality Rendering","papers":132,"papersByYear":[5,null,9,null,41,null,77],"rel":5.1294,"reviewed":true,"share":[0.4651,null,0.5583,null,1.9017,null,2.8508],"size":183,"slope":0.00425},{"claims":[47,null,25,null,68,null,31],"deltaPp":-1.9933,"id":37,"keywords":["shape","deformation","primitive","implicit","3d"],"label":"Implicit Primitive Shape Modeling","labelFull":"Implicit Shape Modeling with Primitive Priors and Deformation","papers":99,"papersByYear":[27,null,17,null,41,null,14],"rel":-0.7936,"reviewed":true,"share":[2.5116,null,1.0546,null,1.9017,null,0.5183],"size":171,"slope":-0.002566},{"claims":[1,null,1,null,53,null,115],"deltaPp":3.128,"id":38,"keywords":["diffusion","text-to-image","latent","video","pre-trained"],"label":"Controllable Text-to-Image Diffusion","labelFull":"Control and Adaptation of Text-to-Image Diffusion Models","papers":130,"papersByYear":[1,null,1,null,41,null,87],"rel":33.6344,"reviewed":true,"share":[0.093,null,0.062,null,1.9017,null,3.221],"size":170,"slope":0.005612},{"claims":[12,null,33,null,46,null,77],"deltaPp":0.7187,"id":39,"keywords":["navigation","vision-and-language","embodied","agents","environments"],"label":"Vision-and-Language Navigation","labelFull":"Vision-and-Language Navigation for Embodied Agents","papers":84,"papersByYear":[7,null,17,null,23,null,37],"rel":1.1037,"reviewed":true,"share":[0.6512,null,1.0546,null,1.0668,null,1.3699],"size":168,"slope":0.001084},{"claims":[23,null,40,null,46,null,56],"deltaPp":0.3295,"id":40,"keywords":["quantization","quantized","mixed-precision","error","weights"],"label":"Mixed-Precision Quantization","labelFull":"Mixed-Precision Quantization for Efficient Inference","papers":70,"papersByYear":[8,null,14,null,19,null,29],"rel":0.4428,"reviewed":true,"share":[0.7442,null,0.8685,null,0.8813,null,1.0737],"size":165,"slope":0.000501},{"claims":[36,null,72,null,38,null,19],"deltaPp":-1.3801,"id":41,"keywords":["loss","class","margin","contrastive","metric learning"],"label":"Class-Balanced Metric Learning","labelFull":"Class-Balanced Metric Learning for Imbalanced Classification","papers":123,"papersByYear":[22,null,50,null,33,null,18],"rel":-0.6744,"reviewed":true,"share":[2.0465,null,3.1017,null,1.5306,null,0.6664],"size":165,"slope":-0.002856},{"claims":[84,null,55,null,15,null,10],"deltaPp":-2.5315,"id":42,"keywords":["saliency","salient","object detection","video object","network"],"label":"Salient Object Detection","labelFull":"Salient Object Detection in Images, Video, and RGB‑D","papers":68,"papersByYear":[30,null,22,null,9,null,7],"rel":-0.9071,"reviewed":true,"share":[2.7907,null,1.3648,null,0.4174,null,0.2592],"size":164,"slope":-0.004271},{"claims":[14,null,44,null,66,null,38],"deltaPp":-0.0227,"id":43,"keywords":["3d object","object detection","3d detection","monocular","detectors"],"label":"Image-based 3D Object Detection","labelFull":"Image-based 3D Object Detection for Monocular and Multi-View","papers":101,"papersByYear":[9,null,25,null,45,null,22],"rel":-0.0271,"reviewed":true,"share":[0.8372,null,1.5509,null,2.0872,null,0.8145],"size":162,"slope":0.000234},{"claims":[0,null,7,null,45,null,105],"deltaPp":2.4806,"id":44,"keywords":["token","attention","merging","pruning","information"],"label":"Token Sparsification","labelFull":"Token Pruning, Merging, Reduction for Efficient Transformers","papers":102,"papersByYear":[0,null,6,null,29,null,67],"rel":null,"reviewed":true,"share":[0.0,null,0.3722,null,1.3451,null,2.4806],"size":157,"slope":0.004207},{"claims":[10,null,37,null,58,null,52],"deltaPp":0.9048,"id":45,"keywords":["student","teacher","continual learning","rehearsal","curriculum"],"label":"Teacher-Student Continual Learning","labelFull":"Teacher-Student Knowledge Distillation for Continual Learning","papers":101,"papersByYear":[5,null,23,null,36,null,37],"rel":1.9454,"reviewed":true,"share":[0.4651,null,1.4268,null,1.6698,null,1.3699],"size":157,"slope":0.001479},{"claims":[26,null,51,null,55,null,22],"deltaPp":-0.5229,"id":46,"keywords":["hand","pose","contact","interacting","estimation"],"label":"3D Hand Pose Estimation","labelFull":"3D Hand Pose and Shape Estimation from Single Color Images","papers":59,"papersByYear":[10,null,19,null,19,null,11],"rel":-0.5621,"reviewed":true,"share":[0.9302,null,1.1787,null,0.8813,null,0.4073],"size":154,"slope":-0.000933},{"claims":[31,null,40,null,44,null,38],"deltaPp":-0.9738,"id":47,"keywords":["attention","multi-head","mechanism","maps","module"],"label":"Attention Mechanism Variants","labelFull":"Efficient and Interpretable Attention Mechanism Variants","papers":119,"papersByYear":[24,null,28,null,33,null,34],"rel":-0.4362,"reviewed":true,"share":[2.2326,null,1.737,null,1.5306,null,1.2588],"size":153,"slope":-0.001564},{"claims":[46,null,60,null,34,null,10],"deltaPp":-1.8803,"id":48,"keywords":["gan","latent","images","inversion","generator"],"label":"GAN Latent Space Structuring","labelFull":"GAN Latent Space Structuring and Regularization","papers":80,"papersByYear":[23,null,29,null,21,null,7],"rel":-0.8789,"reviewed":true,"share":[2.1395,null,1.799,null,0.974,null,0.2592],"size":150,"slope":-0.003233},{"claims":[1,null,0,null,3,null,144],"deltaPp":3.2021,"id":49,"keywords":["multimodal","multi-modal","capabilities","reasoning","understanding"],"label":"Multimodal LLM Benchmarks","labelFull":"Benchmarking and Performance Evaluation of Multimodal LLMs","papers":92,"papersByYear":[1,null,0,null,2,null,89],"rel":34.4312,"reviewed":true,"share":[0.093,null,0.0,null,0.0928,null,3.2951],"size":148,"slope":0.00485},{"claims":[28,null,44,null,42,null,31],"deltaPp":-0.5808,"id":50,"keywords":["super-resolution","blind","low-resolution","image","perceptual"],"label":"Blind Super-Resolution","labelFull":"Blind Image and Video Super-Resolution","papers":90,"papersByYear":[15,null,25,null,28,null,22],"rel":-0.4163,"reviewed":true,"share":[1.3953,null,1.5509,null,1.2987,null,0.8145],"size":145,"slope":-0.000997},{"claims":[19,null,26,null,64,null,34],"deltaPp":-0.3758,"id":51,"keywords":["object segmentation","video object","video instance","instance segmentation","referring"],"label":"Video Object Segmentation","labelFull":"Video Object Segmentation and Instance Segmentation","papers":79,"papersByYear":[12,null,17,null,30,null,20],"rel":-0.3366,"reviewed":true,"share":[1.1163,null,1.0546,null,1.3915,null,0.7405],"size":143,"slope":-0.000395},{"claims":[23,null,56,null,40,null,18],"deltaPp":-1.2311,"id":52,"keywords":["imagenet","top-1","accuracy","resnet-50","higher"],"label":"Efficient ImageNet Classification","labelFull":"Efficient ImageNet Classification and Training","papers":124,"papersByYear":[20,null,52,null,35,null,17],"rel":-0.6617,"reviewed":true,"share":[1.8605,null,3.2258,null,1.6234,null,0.6294],"size":137,"slope":-0.002648},{"claims":[0,null,35,null,91,null,11],"deltaPp":0.3332,"id":53,"keywords":["radiance","fields","neural","rendering","scenes"],"label":"Real-Time Neural Radiance Fields","labelFull":"Real-Time Neural Radiance Fields: Fast Rendering and Editing","papers":97,"papersByYear":[0,null,21,null,67,null,9],"rel":null,"reviewed":true,"share":[0.0,null,1.3027,null,3.1076,null,0.3332],"size":137,"slope":0.001402},{"claims":[0,null,9,null,63,null,62],"deltaPp":1.0737,"id":54,"keywords":["federated","client","learning","privacy","communication"],"label":"Personalized Federated Learning","labelFull":"Personalized Federated Learning for Heterogeneous Clients","papers":61,"papersByYear":[0,null,3,null,29,null,29],"rel":null,"reviewed":true,"share":[0.0,null,0.1861,null,1.3451,null,1.0737],"size":134,"slope":0.00219},{"claims":[6,null,42,null,33,null,51],"deltaPp":0.7946,"id":55,"keywords":["trajectory","prediction","forecasting","future","multi-agent"],"label":"Multi-Agent Trajectory Prediction","labelFull":"Multi-Agent Trajectory Prediction and Motion Forecasting","papers":73,"papersByYear":[3,null,24,null,17,null,29],"rel":2.847,"reviewed":true,"share":[0.2791,null,1.4888,null,0.7885,null,1.0737],"size":132,"slope":0.000842},{"claims":[38,null,47,null,25,null,20],"deltaPp":-2.3664,"id":56,"keywords":["image classification","receptive","feature","multi-scale","pyramid"],"label":"Multi-Scale Pyramid Features","labelFull":"Pyramid Multi-Scale Features for Image Classification","papers":121,"papersByYear":[33,null,44,null,25,null,19],"rel":-0.7709,"reviewed":true,"share":[3.0698,null,2.7295,null,1.1596,null,0.7034],"size":130,"slope":-0.004335},{"claims":[17,null,26,null,33,null,52],"deltaPp":0.3087,"id":57,"keywords":["reconstruction","hdr","quality","ct","compared"],"label":"Deep Image Reconstruction","labelFull":"Deep Learning for Image Reconstruction","papers":99,"papersByYear":[13,null,17,null,28,null,41],"rel":0.2553,"reviewed":true,"share":[1.2093,null,1.0546,null,1.2987,null,1.518],"size":128,"slope":0.000585},{"claims":[7,null,12,null,20,null,88],"deltaPp":1.7003,"id":58,"keywords":["reasoning","explanations","understanding","spatial","visual"],"label":"Visual Chain-of-Thought Reasoning","labelFull":"Chain-of-Thought Reasoning for Vision and Video Understanding","papers":91,"papersByYear":[6,null,11,null,13,null,61],"rel":3.0466,"reviewed":true,"share":[0.5581,null,0.6824,null,0.603,null,2.2584],"size":127,"slope":0.002511},{"claims":[24,null,35,null,31,null,36],"deltaPp":-0.3388,"id":59,"keywords":["illumination","lighting","reflectance","decomposition","captured"],"label":"Multi-Illuminant Intrinsic Image Decomposition","labelFull":"Intrinsic Image Decomposition for Multi-Illuminant Scenes","papers":76,"papersByYear":[12,null,21,null,22,null,21],"rel":-0.3035,"reviewed":true,"share":[1.1163,null,1.3027,null,1.0204,null,0.7775],"size":126,"slope":-0.000649},{"claims":[9,null,18,null,43,null,56],"deltaPp":1.331,"id":60,"keywords":["view synthesis","novel-view","camera","scenes","synthesizes"],"label":"Neural Novel View Synthesis","labelFull":"Neural Rendering for Novel View Synthesis","papers":100,"papersByYear":[4,null,15,null,35,null,46],"rel":3.577,"reviewed":true,"share":[0.3721,null,0.9305,null,1.6234,null,1.7031],"size":126,"slope":0.002343},{"claims":[15,null,37,null,30,null,41],"deltaPp":-0.1916,"id":61,"keywords":["visual","embeddings","relationship","representations","self-supervised"],"label":"Self-Supervised Visual Representations","labelFull":"Self-Supervised Learning of Fine-Grained Visual Representations","papers":95,"papersByYear":[14,null,27,null,24,null,30],"rel":-0.1471,"reviewed":true,"share":[1.3023,null,1.6749,null,1.1132,null,1.1107],"size":123,"slope":-0.000568},{"claims":[11,null,26,null,54,null,28],"deltaPp":0.1254,"id":62,"keywords":["uncertainty","robustness","distribution shifts","calibration","out-of-distribution"],"label":"Uncertainty-Aware Robustness","labelFull":"Uncertainty Estimation for Robustness to Distribution Shifts","papers":100,"papersByYear":[9,null,21,null,44,null,26],"rel":0.1498,"reviewed":true,"share":[0.8372,null,1.3027,null,2.0408,null,0.9626],"size":119,"slope":0.000557},{"claims":[1,null,47,null,41,null,27],"deltaPp":0.6104,"id":63,"keywords":["vision","transformers","backbone","image classification","vanilla"],"label":"Efficient Vision Transformers","labelFull":"Efficient Vision Transformers and Hybrid Backbones","papers":74,"papersByYear":[1,null,26,null,28,null,19],"rel":6.5634,"reviewed":true,"share":[0.093,null,1.6129,null,1.2987,null,0.7034],"size":116,"slope":0.000759},{"claims":[33,null,20,null,23,null,38],"deltaPp":-0.4897,"id":64,"keywords":["scene completion","indoor","geometric","cad","semantic scene"],"label":"3D Scene Completion","labelFull":"3D Scene Completion and Semantic Understanding","papers":83,"papersByYear":[18,null,16,null,17,null,32],"rel":-0.2925,"reviewed":true,"share":[1.6744,null,0.9926,null,0.7885,null,1.1847],"size":114,"slope":-0.000837},{"claims":[8,null,19,null,48,null,36],"deltaPp":0.4595,"id":65,"keywords":["query","decoder","cross-attention","key","self-attention"],"label":"Query-Based Object Detection","labelFull":"Advances in Query-Based Object Detection with Transformers","papers":87,"papersByYear":[7,null,12,null,38,null,30],"rel":0.7056,"reviewed":true,"share":[0.6512,null,0.7444,null,1.7625,null,1.1107],"size":111,"slope":0.001198},{"claims":[17,null,21,null,28,null,41],"deltaPp":-0.1356,"id":66,"keywords":["localization","visual","indoor","cross-view","outdoor"],"label":"Visual Localization","labelFull":"Visual Localization for Indoor, Outdoor, and Cross-View","papers":77,"papersByYear":[13,null,14,null,21,null,29],"rel":-0.1121,"reviewed":true,"share":[1.2093,null,0.8685,null,0.974,null,1.0737],"size":107,"slope":-0.000151},{"claims":[3,null,10,null,30,null,62],"deltaPp":1.498,"id":67,"keywords":["editing","generation","3d","scene","4d"],"label":"3D Scene Generation","labelFull":"Neural 3D/4D Scene Generation and Editing","papers":86,"papersByYear":[3,null,10,null,25,null,48],"rel":5.3673,"reviewed":true,"share":[0.2791,null,0.6203,null,1.1596,null,1.7771],"size":105,"slope":0.002517},{"claims":[8,null,19,null,27,null,46],"deltaPp":0.7007,"id":68,"keywords":["reinforcement learning","agents","reward","policy","imitation"],"label":"Multi-Agent Reinforcement Learning","labelFull":"Multi-Agent Reinforcement Learning for Perception & Planning","papers":69,"papersByYear":[6,null,13,null,16,null,34],"rel":1.2555,"reviewed":true,"share":[0.5581,null,0.8065,null,0.7421,null,1.2588],"size":100,"slope":0.001019},{"claims":[1,null,5,null,30,null,60],"deltaPp":0.9437,"id":69,"keywords":["avatars","hair","animatable","head","human"],"label":"Animatable Human Avatars","labelFull":"Neural 3D Human Avatar Reconstruction for Animation","papers":49,"papersByYear":[1,null,4,null,16,null,28],"rel":10.1473,"reviewed":true,"share":[0.093,null,0.2481,null,0.7421,null,1.0367],"size":96,"slope":0.001663},{"claims":[6,null,7,null,12,null,70],"deltaPp":1.4782,"id":70,"keywords":["gaussian","splatting","reconstruction","3d","single-view"],"label":"3D Gaussian Splatting","labelFull":"3D Gaussian Splatting for High-Fidelity 3D Reconstruction","papers":79,"papersByYear":[6,null,7,null,11,null,55],"rel":2.6486,"reviewed":true,"share":[0.5581,null,0.4342,null,0.5102,null,2.0363],"size":95,"slope":0.002255},{"claims":[11,null,13,null,40,null,29],"deltaPp":-0.1889,"id":71,"keywords":["imaging","compressive","snapshot","phase","single-photon"],"label":"Learned Snapshot Compressive Imaging","labelFull":"Learned Optical Coding for Snapshot Compressive Imaging","papers":48,"papersByYear":[8,null,9,null,16,null,15],"rel":-0.2538,"reviewed":true,"share":[0.7442,null,0.5583,null,0.7421,null,0.5553],"size":93,"slope":-0.000191},{"claims":[6,null,14,null,31,null,42],"deltaPp":0.9228,"id":72,"keywords":["computational","overhead","minimal","memory","modules"],"label":"Efficient Neural Networks","labelFull":"Efficient Neural Networks: Reducing Compute, Memory, Latency","papers":87,"papersByYear":[6,null,14,null,27,null,40],"rel":1.6535,"reviewed":true,"share":[0.5581,null,0.8685,null,1.2523,null,1.4809],"size":93,"slope":0.001576},{"claims":[24,null,32,null,17,null,18],"deltaPp":-1.0631,"id":73,"keywords":["graph","matching","node","edges","networks"],"label":"Neural Graph Matching","labelFull":"Neural Graph Matching: GNN and Hypergraph Techniques","papers":64,"papersByYear":[17,null,21,null,12,null,14],"rel":-0.6723,"reviewed":true,"share":[1.5814,null,1.3027,null,0.5566,null,0.5183],"size":91,"slope":-0.001968},{"claims":[13,null,12,null,23,null,42],"deltaPp":0.4017,"id":74,"keywords":["inference","speed","faster","time","memory"],"label":"Fast Inference","labelFull":"Fast Inference and Memory-Efficient Vision Techniques","papers":85,"papersByYear":[12,null,11,null,21,null,41],"rel":0.3598,"reviewed":true,"share":[1.1163,null,0.6824,null,0.974,null,1.518],"size":90,"slope":0.000748},{"claims":[3,null,18,null,19,null,49],"deltaPp":0.7026,"id":75,"keywords":["restoration","degradation","image","real-world","blind"],"label":"Real-World Blind Image Restoration","labelFull":"Real-World Blind Image Restoration with Degradation Modeling","papers":54,"papersByYear":[2,null,13,null,15,null,24],"rel":3.7774,"reviewed":true,"share":[0.186,null,0.8065,null,0.6957,null,0.8886],"size":89,"slope":0.000999},{"claims":[1,null,8,null,35,null,43],"deltaPp":1.2769,"id":76,"keywords":["text","embeddings","alignment","clip","visual"],"label":"Text-to-Image Alignment","labelFull":"Efficient Alignment and Adaptation of Text-to-Image Models","papers":74,"papersByYear":[1,null,8,null,28,null,37],"rel":13.7301,"reviewed":true,"share":[0.093,null,0.4963,null,1.2987,null,1.3699],"size":87,"slope":0.002317},{"claims":[34,null,17,null,26,null,8],"deltaPp":-1.0802,"id":77,"keywords":["clothing","body","garment","try-on","human"],"label":"3D Clothed Human Reconstruction","labelFull":"3D Reconstruction of Clothed Humans for Virtual Try-On","papers":44,"papersByYear":[14,null,11,null,13,null,6],"rel":-0.8295,"reviewed":true,"share":[1.3023,null,0.6824,null,0.603,null,0.2221],"size":85,"slope":-0.00166},{"claims":[6,null,40,null,20,null,18],"deltaPp":0.0352,"id":78,"keywords":["scene graph","graph generation","genome","context","relation"],"label":"Scene Graph Generation","labelFull":"Scene Graph Generation for Visual Relationship Understanding","papers":44,"papersByYear":[4,null,19,null,10,null,11],"rel":0.0946,"reviewed":true,"share":[0.3721,null,1.1787,null,0.4638,null,0.4073],"size":84,"slope":-0.000305},{"claims":[32,null,12,null,24,null,16],"deltaPp":-0.56,"id":79,"keywords":["text","recognition","scene","chinese","character"],"label":"Scene Text Recognition","labelFull":"Scene Text Recognition: Datasets, Benchmarks, Robust Models","papers":35,"papersByYear":[10,null,6,null,9,null,10],"rel":-0.602,"reviewed":true,"share":[0.9302,null,0.3722,null,0.4174,null,0.3702],"size":84,"slope":-0.000817},{"claims":[2,null,11,null,17,null,52],"deltaPp":0.7396,"id":80,"keywords":["remote sensing","satellite","foundation","imagery","context"],"label":"Geospatial Foundation Models","labelFull":"Geospatial Foundation Models for Remote Sensing","papers":44,"papersByYear":[2,null,6,null,11,null,25],"rel":3.9763,"reviewed":true,"share":[0.186,null,0.3722,null,0.5102,null,0.9256],"size":82,"slope":0.001178},{"claims":[2,null,16,null,23,null,40],"deltaPp":0.3323,"id":81,"keywords":["watermark","lossless","image","robustness","embeds"],"label":"Robust Image Watermarking","labelFull":"Robust Deep Learning-based Invisible Image Watermarking","papers":33,"papersByYear":[2,null,7,null,10,null,14],"rel":1.7866,"reviewed":true,"share":[0.186,null,0.4342,null,0.4638,null,0.5183],"size":81,"slope":0.000513},{"claims":[14,null,6,null,22,null,39],"deltaPp":-0.1347,"id":82,"keywords":["mesh","surfaces","topology","implicit","representation"],"label":"Neural Mesh Generation","labelFull":"Neural Mesh Generation and Reconstruction","papers":59,"papersByYear":[11,null,4,null,20,null,24],"rel":-0.1316,"reviewed":true,"share":[1.0233,null,0.2481,null,0.9276,null,0.8886],"size":81,"slope":0.000138},{"claims":[16,null,17,null,26,null,21],"deltaPp":-0.7289,"id":83,"keywords":["occlusion","viewpoint","views","attention","objects"],"label":"Occlusion-Aware Multi-View Learning","labelFull":"Occlusion-Aware Multi-View Learning for Robust Vision","papers":71,"papersByYear":[15,null,15,null,23,null,18],"rel":-0.5224,"reviewed":true,"share":[1.3953,null,0.9305,null,1.0668,null,0.6664],"size":80,"slope":-0.001025},{"claims":[1,null,6,null,18,null,51],"deltaPp":0.8696,"id":84,"keywords":["occupancy","prediction","3d","representation","nuscenes"],"label":"3D Occupancy Prediction","labelFull":"3D Occupancy Prediction and Forecasting for Autonomous Driving","papers":38,"papersByYear":[1,null,4,null,7,null,26],"rel":9.3505,"reviewed":true,"share":[0.093,null,0.2481,null,0.3247,null,0.9626],"size":76,"slope":0.001343},{"claims":[15,null,22,null,19,null,20],"deltaPp":-0.579,"id":85,"keywords":["mesh","human","recovery","body","3d"],"label":"3D Human Mesh Recovery","labelFull":"3D Human Mesh Recovery from Single and Multi-View Images","papers":53,"papersByYear":[11,null,15,null,15,null,12],"rel":-0.5658,"reviewed":true,"share":[1.0233,null,0.9305,null,0.6957,null,0.4443],"size":76,"slope":-0.000986},{"claims":[10,null,18,null,23,null,25],"deltaPp":-0.2629,"id":86,"keywords":["denoising","noise","restoration","noisy","image"],"label":"Self-Supervised Image Denoising","labelFull":"Self-Supervised and Unsupervised Image Denoising","papers":46,"papersByYear":[8,null,11,null,14,null,13],"rel":-0.3533,"reviewed":true,"share":[0.7442,null,0.6824,null,0.6494,null,0.4813],"size":76,"slope":-0.000411},{"claims":[16,null,15,null,27,null,17],"deltaPp":-0.28,"id":87,"keywords":["lane","detection","bev","road","line"],"label":"3D Lane Detection","labelFull":"3D Lane Detection and Bird's-Eye-View Lane Graph Estimation","papers":23,"papersByYear":[5,null,5,null,8,null,5],"rel":-0.602,"reviewed":true,"share":[0.4651,null,0.3102,null,0.3711,null,0.1851],"size":75,"slope":-0.00039},{"claims":[0,null,2,null,8,null,63],"deltaPp":1.518,"id":88,"keywords":["tokens","visual","autoregressive","spectral","pruning"],"label":"Visual Token Reduction","labelFull":"Visual Token Reduction for Efficient Vision Transformers","papers":50,"papersByYear":[0,null,1,null,8,null,41],"rel":null,"reviewed":true,"share":[0.0,null,0.062,null,0.3711,null,1.518],"size":73,"slope":0.002432},{"claims":[7,null,8,null,18,null,39],"deltaPp":0.6266,"id":89,"keywords":["generative","latent","conditional","score-based","introduced"],"label":"Generative Model Evaluation","labelFull":"Evaluation, Benchmarking, and Fingerprinting of Generative Models","papers":58,"papersByYear":[6,null,7,null,13,null,32],"rel":1.1227,"reviewed":true,"share":[0.5581,null,0.4342,null,0.603,null,1.1847],"size":72,"slope":0.001024},{"claims":[18,null,18,null,18,null,16],"deltaPp":-0.411,"id":90,"keywords":["slam","odometry","camera","tracking","eye"],"label":"Neural SLAM","labelFull":"Neural SLAM: Deep Learning for Visual Localization & Mapping","papers":36,"papersByYear":[8,null,10,null,9,null,9],"rel":-0.5523,"reviewed":true,"share":[0.7442,null,0.6203,null,0.4174,null,0.3332],"size":70,"slope":-0.000718},{"claims":[19,null,18,null,11,null,19],"deltaPp":-0.355,"id":91,"keywords":["gaze","eye","target","cross-dataset","estimation"],"label":"Generalizable Gaze Estimation","labelFull":"Cross-Dataset Generalizable Gaze Estimation and Tracking","papers":24,"papersByYear":[7,null,6,null,3,null,8],"rel":-0.5451,"reviewed":true,"share":[0.6512,null,0.3722,null,0.1391,null,0.2962],"size":67,"slope":-0.000649},{"claims":[1,null,15,null,36,null,14],"deltaPp":0.0921,"id":92,"keywords":["spiking","neural networks","time","conversion","mamba"],"label":"Low-Latency Spiking Neural Networks","labelFull":"Low-Latency Efficient Spiking Neural Networks","papers":25,"papersByYear":[1,null,6,null,13,null,5],"rel":0.9903,"reviewed":true,"share":[0.093,null,0.3722,null,0.603,null,0.1851],"size":66,"slope":0.000254},{"claims":[1,null,3,null,14,null,48],"deltaPp":1.536,"id":93,"keywords":["diffusion","3d","2d","rendered","normals"],"label":"3D Diffusion Models","labelFull":"Diffusion Models for 3D Object Generation and Reconstruction","papers":56,"papersByYear":[1,null,1,null,10,null,44],"rel":16.5161,"reviewed":true,"share":[0.093,null,0.062,null,0.4638,null,1.629],"size":66,"slope":0.002505},{"claims":[1,null,16,null,25,null,22],"deltaPp":0.3883,"id":94,"keywords":["lidar","3d object","single-photon","calibration","object detection"],"label":"LiDAR 3D Detection","labelFull":"LiDAR-based 3D Object Detection, Fusion, and Adaptation","papers":40,"papersByYear":[1,null,12,null,14,null,13],"rel":4.1753,"reviewed":true,"share":[0.093,null,0.7444,null,0.6494,null,0.4813],"size":64,"slope":0.000535},{"claims":[5,null,17,null,21,null,20],"deltaPp":0.4804,"id":95,"keywords":["frequency","components","amplitude","spectrum","fourier"],"label":"Fourier Domain Learning","labelFull":"Fourier Domain Learning for Vision Models","papers":44,"papersByYear":[2,null,10,null,14,null,18],"rel":2.5828,"reviewed":true,"share":[0.186,null,0.6203,null,0.6494,null,0.6664],"size":63,"slope":0.000735},{"claims":[9,null,12,null,19,null,23],"deltaPp":-0.1138,"id":96,"keywords":["layout","generation","elements","planning","bounding"],"label":"Layout Generation","labelFull":"Generative Layout Synthesis for Images, Documents, and UI","papers":31,"papersByYear":[6,null,4,null,9,null,12],"rel":-0.2039,"reviewed":true,"share":[0.5581,null,0.2481,null,0.4174,null,0.4443],"size":63,"slope":-8.6e-05},{"claims":[16,null,15,null,22,null,9],"deltaPp":-1.1552,"id":97,"keywords":["cifar-10","cifar-100","imagenet","cifar10","benchmarks"],"label":"Image Classification Benchmarks","labelFull":"Standard Image Classification Benchmarks (CIFAR, ImageNet)","papers":58,"papersByYear":[16,null,14,null,19,null,9],"rel":-0.7761,"reviewed":true,"share":[1.4884,null,0.8685,null,0.8813,null,0.3332],"size":62,"slope":-0.001726},{"claims":[25,null,18,null,16,null,2],"deltaPp":-1.1353,"id":98,"keywords":["crowd","counting","density","annotations","semi-supervised"],"label":"Crowd Counting","labelFull":"Deep Learning for Crowd Counting and Density Map Generation","papers":29,"papersByYear":[13,null,9,null,5,null,2],"rel":-0.9388,"reviewed":true,"share":[1.2093,null,0.5583,null,0.2319,null,0.074],"size":61,"slope":-0.001866},{"claims":[2,null,9,null,18,null,31],"deltaPp":0.4064,"id":99,"keywords":["enhancement","low-light","image","illumination","event"],"label":"Low-Light Image Enhancement","labelFull":"Low-Light Image and Video Enhancement","papers":36,"papersByYear":[2,null,6,null,12,null,16],"rel":2.1849,"reviewed":true,"share":[0.186,null,0.3722,null,0.5566,null,0.5924],"size":60,"slope":0.000702},{"claims":[4,null,2,null,2,null,51],"deltaPp":1.0167,"id":100,"keywords":["video generation","generation quality","test-time","multi-task","variational"],"label":"Video Diffusion Models","labelFull":"Diffusion Video Generation: Quality, Efficiency, and Control","papers":42,"papersByYear":[3,null,2,null,2,null,35],"rel":3.6428,"reviewed":true,"share":[0.2791,null,0.1241,null,0.0928,null,1.2958],"size":59,"slope":0.001509},{"claims":[0,null,4,null,17,null,37],"deltaPp":1.2218,"id":101,"keywords":["diffusion","denoising","timesteps","process","priors"],"label":"Diffusion Priors","labelFull":"Efficient Diffusion Prior Utilization","papers":49,"papersByYear":[0,null,2,null,14,null,33],"rel":null,"reviewed":true,"share":[0.0,null,0.1241,null,0.6494,null,1.2218],"size":58,"slope":0.002095},{"claims":[10,null,6,null,17,null,25],"deltaPp":0.1273,"id":102,"keywords":["referring","expression","comprehension","image segmentation","linguistic"],"label":"Referring Expression Grounding","labelFull":"Referring Expression Grounding and Segmentation","papers":33,"papersByYear":[5,null,3,null,9,null,16],"rel":0.2737,"reviewed":true,"share":[0.4651,null,0.1861,null,0.4174,null,0.5924],"size":58,"slope":0.000307},{"claims":[0,null,5,null,10,null,42],"deltaPp":1.1847,"id":103,"keywords":["video understanding","video-language","long","streaming","state-of-the-art multiple"],"label":"Long Video Understanding","labelFull":"Long Video Understanding with Multimodal LLMs","papers":43,"papersByYear":[0,null,5,null,6,null,32],"rel":null,"reviewed":true,"share":[0.0,null,0.3102,null,0.2783,null,1.1847],"size":57,"slope":0.001761}],"totals":{"claims":[5306,0,7989,0,10695,0,13820],"papers":[1075,0,1612,0,2156,0,2701]},"years":[2019,2020,2021,2022,2023,2024,2025]}