List your company
ResearchFormerly WizWand

Visual Question Answering on CLEVR-Humans 1.0 (test)

85.5AccuracyIPRM
25.1840.8456.572.16May 10, 2017Aug 11, 2018Nov 13, 2019Feb 13, 2021May 18, 2022Aug 19, 2023Nov 20, 2024
Updated 8mo ago

Evaluation results

MethodLinks
IPRMEvaluation Protocol=Fine-tuning2024.1185.5
MDETREvaluation Protocol=Fine-tuning, Supervision=Object bounding-boxes2024.1181.7
MACEvaluation Protocol=Fine-tuning2024.1181.5
LEFTEvaluation Protocol=Fine-tuning2024.1178.8
FiLMEvaluation Protocol=Fine-tuning2024.1175.9
GLTEvaluation Protocol=Fine-tuning2024.1175.8
RAMENEvaluation Protocol=Fine-tuning2024.1167.8
IEPTraining Protocol=Train CLEVR, finetune human, Ground-truth programs=18K2017.0566.6
PG+EEEvaluation Protocol=Fine-tuning, Supervision=Functional programs / pre-defined dataset-specific neural modules2024.1166.6
IPRMEvaluation Protocol=Zero-shot2024.1163.8
MDETREvaluation Protocol=Zero-shot, Supervision=Object bounding-boxes2024.1159.9
NS-VQAEvaluation Protocol=Zero-shot, Supervision=Functional programs / pre-defined dataset-specific neural modules and object bounding-boxes2024.1157.8
CNN+LSTM+SA+MLPTraining Protocol=Train CLEVR, finetune human2017.0557.6
MACEvaluation Protocol=Zero-shot2024.1157.4
FiLMEvaluation Protocol=Zero-shot2024.1156.6
IEPTraining Protocol=Train CLEVR, Ground-truth programs=18K2017.0554
PG+EEEvaluation Protocol=Zero-shot, Supervision=Functional programs / pre-defined dataset-specific neural modules2024.1154
CNN+LSTM+SA+MLPTraining Protocol=Train CLEVR2017.0550.4
CNN+LSTMTraining Protocol=Train CLEVR, finetune human2017.0543.2
CNN+LSTMTraining Protocol=Train CLEVR2017.0537.7
LSTMTraining Protocol=Train CLEVR, finetune human2017.0536.5
LSTMTraining Protocol=Train CLEVR2017.0527.5