fastreid.evaluation.roc Namespace Reference

Functions
	evaluate_roc_py (distmat, q_feats, g_feats, q_pids, g_pids, q_camids, g_camids)
	evaluate_roc (distmat, q_feats, g_feats, q_pids, g_pids, q_camids, g_camids, use_cython=True)

Variables
bool	IS_CYTHON_AVAI = True

Detailed Description

@author:  l1aoxingyu
@contact: sherlockliao01@gmail.com

Function Documentation

evaluate_roc()

fastreid.evaluation.roc.evaluate_roc	(		distmat,
			q_feats,
			g_feats,
			q_pids,
			g_pids,
			q_camids,
			g_camids,
			use_cython = True )

Evaluates CMC rank.
Args:
    distmat (numpy.ndarray): distance matrix of shape (num_query, num_gallery).
    q_pids (numpy.ndarray): 1-D array containing person identities
        of each query instance.
    g_pids (numpy.ndarray): 1-D array containing person identities
        of each gallery instance.
    q_camids (numpy.ndarray): 1-D array containing camera views under
        which each query instance is captured.
    g_camids (numpy.ndarray): 1-D array containing camera views under
        which each gallery instance is captured.
    use_cython (bool, optional): use cython code for evaluation. Default is True.
        This is highly recommended as the cython code can speed up the cmc computation
        by more than 10x. This requires Cython to be installed.

Definition at line 68 of file roc.py.

):
    """Evaluates CMC rank.
    Args:
        distmat (numpy.ndarray): distance matrix of shape (num_query, num_gallery).
        q_pids (numpy.ndarray): 1-D array containing person identities
            of each query instance.
        g_pids (numpy.ndarray): 1-D array containing person identities
            of each gallery instance.
        q_camids (numpy.ndarray): 1-D array containing camera views under
            which each query instance is captured.
        g_camids (numpy.ndarray): 1-D array containing camera views under
            which each gallery instance is captured.
        use_cython (bool, optional): use cython code for evaluation. Default is True.
            This is highly recommended as the cython code can speed up the cmc computation
            by more than 10x. This requires Cython to be installed.
    """
    if use_cython and IS_CYTHON_AVAI:
        return evaluate_roc_cy(distmat, q_feats, g_feats, q_pids, g_pids, q_camids, g_camids)
    else:
        return evaluate_roc_py(distmat, q_feats, g_feats, q_pids, g_pids, q_camids, g_camids)

evaluate_roc_py()

fastreid.evaluation.roc.evaluate_roc_py	(		distmat,
			q_feats,
			g_feats,
			q_pids,
			g_pids,
			q_camids,
			g_camids )

Evaluation with ROC curve.
Key: for each query identity, its gallery images from the same camera view are discarded.

Args:
    distmat (np.ndarray): cosine distance matrix

Definition at line 24 of file roc.py.

def evaluate_roc_py(distmat, q_feats, g_feats, q_pids, g_pids, q_camids, g_camids):
    r"""Evaluation with ROC curve.
    Key: for each query identity, its gallery images from the same camera view are discarded.
 
    Args:
        distmat (np.ndarray): cosine distance matrix
    """
    num_q, num_g = distmat.shape
    dim = q_feats.shape[1]
 
    index = faiss.IndexFlatL2(dim)
    index.add(g_feats)
 
    _, indices = index.search(q_feats, k=num_g)
    matches = (g_pids[indices] == q_pids[:, np.newaxis]).astype(np.int32)
 
    pos = []
    neg = []
    for q_idx in range(num_q):
        # get query pid and camid
        q_pid = q_pids[q_idx]
        q_camid = q_camids[q_idx]
 
        # Remove gallery samples that have the same pid and camid with query
        order = indices[q_idx]
        remove = (g_pids[order] == q_pid) & (g_camids[order] == q_camid)
        keep = np.invert(remove)
        raw_cmc = matches[q_idx][keep]
 
        sort_idx = order[keep]
 
        q_dist = distmat[q_idx]
        ind_pos = np.where(raw_cmc == 1)[0]
        pos.extend(q_dist[sort_idx[ind_pos]])
 
        ind_neg = np.where(raw_cmc == 0)[0]
        neg.extend(q_dist[sort_idx[ind_neg]])
 
    scores = np.hstack((pos, neg))
 
    labels = np.hstack((np.zeros(len(pos)), np.ones(len(neg))))
    return scores, labels
 
 

Variable Documentation

IS_CYTHON_AVAI

bool fastreid.evaluation.roc.IS_CYTHON_AVAI = True

Definition at line 15 of file roc.py.