search

yrcong / RelTR

RelTR: Relation Transformer for Scene Graph Generation: https://arxiv.org/abs/2201.11460v2
229 stars 45 forks source link

Train and Test for custom images #14

Closed Chenghao-Ding closed 1 year ago

Chenghao-Ding commented 1 year ago

How did you get the annotations of Visual Genome (in COCO-format)? Do you have tools to process the original datasets? I want to use in my custom datasets(only pictures), I have seen the KaihuaTang/Scene-Graph-Benchmark.pytorch code ,but have no idea to process the my custom datasets for training and testing. Thanks very much!

yrcong commented 1 year ago

We adopt the VG annotations provided by Neural Motifs and convert them into COCO format with the pre-processing code. More details please check #6

Chenghao-Ding commented 1 year ago

We adopt the VG annotations provided by Neural Motifs and convert them into COCO format with the pre-processing code. More details please check #6

Thanks for your timely reply! I find another way to solve my problem that is two-stage way including objects detection and relationships detection. Although two parts are splitted, but for my experiment may have an effective result. Your works looks very nice , thanks for your good job.

rebelzion commented 1 year ago

@Chenghao-Ding did you train on custom images from scratch or fine-tune? How did you fine-tune?