=========================================== The Caltech-UCSD Birds-200-2011 Dataset

Instructions to train model

Get data using http://www.vision.caltech.edu/datasets/cub_200_2011/ Untar using tar -czvf CUB_200_2011.tgz Place the python files from this repo in the same folder.

Execute below code only once

To get the data in one folder and the desired info: python pre_process.py

To load the images and save as np format so we don't have to re-run it for every training iteration Note: this saves the labels in the y_train and y_test data so you might want to change it to use bounding boxes, segmentations. python save_as_np.py

Training the model

Before creating the model if checkpoints are desired, create the folder specified in checkpoint_path variable. python train.py

Below is original readme from the dataset

---

For more information about the dataset, visit the project website:

http://www.vision.caltech.edu/visipedia

If you use the dataset in a publication, please cite the dataset in the style described on the dataset website (see url above).

Directory Information

• images/ The images organized in subdirectories based on species. See IMAGES AND CLASS LABELS section below for more info.
• parts/ 15 part locations per image. See PART LOCATIONS section below for more info.
• attributes/ 322 binary attribute labels from MTurk workers. See ATTRIBUTE LABELS section below for more info.

========================= IMAGES AND CLASS LABELS:

Images are contained in the directory images/, with 200 subdirectories (one for each bird species)

------- List of image files (images.txt) ------ The list of image file names is contained in the file images.txt, with each line corresponding to one image:

<image_id> <image_name>

------- Train/test split (train_test_split.txt) ------ The suggested train/test split is contained in the file train_test_split.txt, with each line corresponding to one image:

<image_id> <is_training_image>

where <image_id> corresponds to the ID in images.txt, and a value of 1 or 0 for <is_training_image> denotes that the file is in the training or test set, respectively.

------- List of class names (classes.txt) ------ The list of class names (bird species) is contained in the file classes.txt, with each line corresponding to one class:

<class_id> <class_name>

------- Image class labels (image_class_labels.txt) ------ The ground truth class labels (bird species labels) for each image are contained in the file image_class_labels.txt, with each line corresponding to one image:

<image_id> <class_id>

where <image_id> and <class_id> correspond to the IDs in images.txt and classes.txt, respectively.

========================= BOUNDING BOXES:

Each image contains a single bounding box label. Bounding box labels are contained in the file bounding_boxes.txt, with each line corresponding to one image:

<image_id>

where <image_id> corresponds to the ID in images.txt, and , , , and are all measured in pixels

========================= PART LOCATIONS:

------- List of part names (parts/parts.txt) ------ The list of all part names is contained in the file parts/parts.txt, with each line corresponding to one part:

<part_id> <part_name>

------- Part locations (parts/part_locs.txt) ------ The set of all ground truth part locations is contained in the file parts/part_locs.txt, with each line corresponding to the annotation of a particular part in a particular image:

<image_id> <part_id>

where <image_id> and <part_id> correspond to the IDs in images.txt and parts/parts.txt, respectively. and denote the pixel location of the center of the part. is 0 if the part is not visible in the image and 1 otherwise.

------- MTurk part locations (parts/part_click_locs.txt) ------ A set of multiple part locations for each image and part, as perceived by multiple MTurk users is contained in parts/part_click_locs.txt, with each line corresponding to the annotation of a particular part in a particular image by a different MTurk worker:

<image_id> <part_id>

where <image_id>, <part_id>, , are in the same format as defined in parts/part_locs.txt, and is the time in seconds spent by the MTurk worker.

========================= ATTRIBUTE LABELS:

------- List of attribute names (attributes/attributes.txt) ------ The list of all attribute names is contained in the file attributes/attributes.txt, with each line corresponding to one attribute:

<attribute_id> <attribute_name>

------- List of certainty names (attributes/certainties.txt) ------ The list of all certainty names (used by workers to specify their certainty of an attribute response of is contained in the file attributes/certainties.txt, with each line corresponding to one certainty:

<certainty_id> <certainty_name>

------- MTurk image attribute labels (attributes/image_attribute_labels.txt) ------ The set of attribute labels as perceived by MTurkers for each image is contained in the file attributes/image_attribute_labels.txt, with each line corresponding to one image/attribute/worker triplet:

<image_id> <attribute_id> <is_present> <certainty_id>

where <image_id>, <attribute_id>, <certainty_id> correspond to the IDs in images.txt, attributes/attributes.txt, and attributes/certainties.txt respectively. <is_present> is 0 or 1 (1 denotes that the attribute is present). denotes the time spent by the MTurker in seconds.

------- Class attribute labels (attributes/class_attribute_labels_continuous.txt) ------ Attributes on a per-class level--in a similar format to the Animals With Attributes dataset--are contained in attributes/class_attribute_labels_continuous.txt. The file contains 200 lines and 312 space-separated columns. Each line corresponds to one class (in the same order as classes.txt) and each column contains one real-valued number corresponding to one attribute (in the same order as attributes.txt). The number is the percentage of the time (between 0 and 100) that a human thinks that the attribute is present for a given class