Makefile

.RECIPEPREFIX +=

ROOT=~/datasets/VQA2
TRAIN_ANN=${ROOT}/v2_mscoco_train2014_annotations.json
TRAIN_QUES=${ROOT}/v2_OpenEnded_mscoco_train2014_questions.json
TRAIN_IMGS=image_embeddings/coco_train_resnet152_pool5.pth

VAL_ANN=${ROOT}/v2_mscoco_val2014_annotations.json
VAL_QUES=${ROOT}/v2_OpenEnded_mscoco_val2014_questions.json
VAL_IMGS=image_embeddings/coco_val_resnet152_pool5.pth

# change this depending on the feature extractor used
# VGG models have dim 4096, Resnet has 2048
IMAGE_DIM=2048

IMAGE_ROOT=~/datasets/coco
ARCH=DeeperLSTM
BATCH=128
WORKERS=8

CHECKPOINT=weights/vqa_checkpoint_DeeperLSTM_200.pth

main: train evaluate

train:
    python main.py $(TRAIN_ANN) $(TRAIN_QUES) $(VAL_ANN) $(VAL_QUES) --images $(TRAIN_IMGS) --val_images $(VAL_IMGS) \
    --arch $(ARCH) --batch_size ${BATCH} --num_workers ${WORKERS} --image_root $(IMAGE_ROOT) --image_dim $(IMAGE_DIM)

train_mcb:
    python main.py $(TRAIN_ANN) $(TRAIN_QUES) $(VAL_ANN) $(VAL_QUES) --images $(TRAIN_IMGS) --val_images $(VAL_IMGS) \
    --arch MCBModel --batch_size 64 --num_workers ${WORKERS} --image_root $(IMAGE_ROOT) --raw_images --img_size 448 --image_dim $(IMAGE_DIM)


raw_images:
    python main.py $(TRAIN_ANN) $(TRAIN_QUES) $(VAL_ANN) $(VAL_QUES) \
    --raw_images --image_root $(IMAGE_ROOT) --arch $(ARCH) --batch_size 32

options:
    python main.py -h

evaluate:
    python evaluate.py $(TRAIN_ANN) $(TRAIN_QUES) $(VAL_ANN) $(VAL_QUES) --images $(TRAIN_IMGS) --val_images $(VAL_IMGS) \
      --batch_size 1 --resume $(CHECKPOINT) --num_workers ${WORKERS} --image_dim $(IMAGE_DIM)

preprocess_train:
    python preprocess_images.py $(IMAGE_ROOT)/annotations/instances_train2014.json --root $(IMAGE_ROOT) --split train --arch resnet152

preprocess_val:
    python preprocess_images.py $(IMAGE_ROOT)/annotations/instances_val2014.json --root $(IMAGE_ROOT) --split val --arch resnet152

demo:
    python demo.py demo_img.jpg "what room is this?" $(TRAIN_QUES) $(TRAIN_ANN) --checkpoint $(CHECKPOINT)