anpana.blogg.se

The train script will dump checkpoints into the folder specified by -checkpoint_path (default = save/). $ python train.py -id fc -caption_model fc -input_json data/cocotalk.json -input_fc_dir data/cocotalk_fc -input_att_dir data/cocotalk_att -input_label_h5 data/cocotalk_label.h5 -batch_size 10 -learning_rate 5e-4 -learning_rate_decay_start 0 -scheduled_sampling_start 0 -checkpoint_path log_fc -save_checkpoint_every 6000 -val_images_use 5000 -max_epochs 30 This file provides preprocessed captions and also standard train-val-test splits. Extract dataset_coco.json from the zip file and copy it in to data/. Train your own network on COCO Download COCO captions and preprocess themĭownload preprocessed coco captions from link from Karpathy's homepage. If you want to do evaluation only, you can then follow this section after downloading the pretrained models (and also the pretrained resnet101). And the performances of each model will be maintained in this issue. Pretrained models (using resnet101 feature) The models can be downloaded from here, and should be placed in data/imagenet_weights. ( Skip if you are using bottom-up feature): If you want to use resnet to extract image features, you need to download pretrained resnet model for both training and evaluation. Python 2.7 (because there is no coco-caption version for python 3)Ĭider (already been added as a submodule) Add transformer (merged from Transformer_captioning).(Evaluation on arbitrary images is not supported.) This is based on my ImageCaptioning.pytorch repository.

The att2in2 model can achieve more than 1.20 Cider score on Karpathy's test split (with self-critical training, bottom-up feature, large rnn hidden size, without ensemble) The author of SCST helped me a lot when I tried to replicate the result. This repository includes the unofficial implementation Self-critical Sequence Training for Image Captioning and Bottom-Up and Top-Down Attention for Image Captioning and Visual Question Answering. Self-critical Sequence Training for Image Captioning (+ misc.)