1e619d91956c813d2328a5460da0a273fe135905,hook/zmes_hook_helpers/face.py,Face,detect,#Face#Any#,123

Before Change

        matched_face_names = []
        matched_face_rects = []

        for idx,face_encoding in enumerate(face_encodings):
            preds = self.svm_model.predict_proba([face_encoding])[0]

            print (preds, self.svm_model.classes_)
            best_pred_ndx = np.argmax(preds)
            best_pred = preds[best_pred_ndx]
            loc = face_locations[idx]

            if best_pred >= g.config["face_recog_min_confidence"]:
                 matched_face_names.append(self.svm_model.classes_[best_pred_ndx])
                 g.logger.debug("face:{} matched with confidence: {}".format(self.svm_model.classes_[best_pred_ndx], best_pred))
            else:     
                g.logger.debug ("face matched:{} but confidence of:{} is less than {}, marking it unknown".format(self.svm_model.classes_[best_pred_ndx], best_pred, g.config["face_recog_min_confidence"]))
                matched_face_names.append(g.config["unknown_face_name"])
                best_pred = 1 // if unknown, don"t carry over pred prob
            matched_face_rects.append((loc[3], loc[0], loc[1], loc[2]))
            conf.append(best_pred)
        return matched_face_rects, matched_face_names, conf

After Change

        // Use the KNN model to find the best matches for the test face
        closest_distances = self.knn.kneighbors(face_encodings, n_neighbors=1)
        are_matches = [closest_distances[0][i][0] <= g.config["face_recog_dist_threshold"] for i in range(len(face_locations))]

        matched_face_names = []
        matched_face_rects = []

Instances