def getting_edge_weights(file_curr, out_dir_labels,out_dir,k, set_k = set_k_mul, normalize_k = normalize_k_mul):
npy_files, anno_all = readTrainTestFile(file_curr)
k_count = np.zeros((len(class_names),k,k))
k_count_big = np.zeros((k,k))
for npy_file,anno_curr in zip(npy_files,anno_all):
label_file = os.path.join(out_dir_labels, os.path.split(npy_file)[1])
labels = np.load(label_file)
k_count_big = set_k(k_count_big,labels)
for gt_idx in np.where(anno_curr)[0]:
k_count[gt_idx] = set_k(k_count[gt_idx],labels)
k_count_big = normalize_k(k_count_big)
print k_count_big.shape
out_file = os.path.join(out_dir,'all_classes_mul.npy')
np.save(out_file, k_count_big )
out_file = os.path.join(out_dir,'all_classes_mul.jpg')
visualize.saveMatAsImage(k_count_big, out_file)
for class_idx in range(len(class_names)):
k_count[class_idx] = normalize_k(k_count[class_idx])
class_name = class_names[class_idx]
out_file = os.path.join(out_dir,class_name+'.npy')
np.save(out_file, k_count[class_idx])
out_file = os.path.join(out_dir,class_name+'.jpg')
visualize.saveMatAsImage(k_count[class_idx], out_file)
visualize.writeHTMLForFolder(out_dir)
def plot_all_necessaries(just_vid_name, out_dir_cooc_viz, out_dir_fg,
out_dir_bg, gt_arr_row, gt_arr_col, arr_cooc):
# just_vid_name = vid_name[:vid_name.rindex('.')]
out_file_cooc = os.path.join(out_dir_cooc_viz, just_vid_name + '.jpg')
out_file_fg = os.path.join(out_dir_fg, just_vid_name + '.jpg')
out_file_bg = os.path.join(out_dir_bg, just_vid_name + '.jpg')
# eye = -2*np.eye(arr_cooc.shape[0])
# arr_cooc_h = arr_cooc+eye
fg_all = arr_cooc[gt_arr_row > 0, :]
fg_fg = fg_all[:, gt_arr_col > 0]
bg_all = arr_cooc[gt_arr_row == 0, :]
bg_bg = bg_all[:, gt_arr_col == 0]
num_bins = np.arange(0, 1.1, .1)
all_vals = [val.flatten() for val in [fg_all, fg_fg, bg_all, bg_bg]]
legend_entries = ['FG All', 'FG FG', 'BG All', 'BG BG']
xlabel = 'Cooc Value'
ylabel = 'Frequency'
xtick_labels = ['%.1f' % val for val in num_bins]
title = 'Foreground Hist for ' + just_vid_name
visualize.plotMultiHist(out_file_fg,
vals=all_vals[:2],
num_bins=[num_bins, num_bins],
legend_entries=legend_entries[:2],
title=title,
xlabel=xlabel,
ylabel=ylabel,
xticks=xtick_labels,
density=True,
align='mid')
title = 'Foreground Hist for ' + just_vid_name
visualize.plotMultiHist(out_file_bg,
vals=all_vals[2:],
num_bins=[num_bins, num_bins],
legend_entries=legend_entries[2:],
title=title,
xlabel=xlabel,
ylabel=ylabel,
xticks=xtick_labels,
density=True,
align='mid')
title = 'Mat ' + just_vid_name
arr_cooc_mat = arr_cooc
arr_cooc_mat[arr_cooc_mat < 0] = 0
visualize.saveMatAsImage(arr_cooc_mat, out_file_cooc, title=title)
print out_file_cooc
def save_graphs_to_look_at(model_file, graph_nums):
out_dir_meta = model_file[:model_file.rindex('.')]
out_dir_meta_meta = out_dir_meta + '_graph_etc'
out_dir_viz = out_dir_meta_meta + '_viz'
util.mkdir(out_dir_viz)
for graph_num in graph_nums:
out_dir_meta = out_dir_meta_meta + '_' + str(graph_num)
assert os.path.exists(out_dir_meta)
vid_files = glob.glob(os.path.join(out_dir_meta, '*test*.npz'))
for vid_file in vid_files:
npz_data = np.load(vid_file)
vid_file = os.path.split(vid_file)[1]
affinity = npz_data['affinity']
gt_vecs = npz_data['gt_vecs']
gt_classes = npz_data['gt_classes']
x_all = npz_data['x_all']
plotter = []
legend_entries = []
for gt_idx, gt_class in enumerate(gt_classes):
gt_vec = gt_vecs[gt_idx]
val_rel = x_all[0, :, gt_class]
gt_vec = gt_vec / np.max(gt_vec)
gt_vec = gt_vec * np.max(val_rel)
# (gt_idx+1)
x_axis = range(gt_vec.size)
plotter.append((x_axis, gt_vec))
plotter.append((x_axis, val_rel))
legend_entries.append(class_names[gt_class])
legend_entries.append(class_names[gt_class] + ' pred')
out_file = os.path.join(
out_dir_viz, vid_file[:vid_file.rindex('.')] + '_gt.jpg')
visualize.plotSimple(plotter,
out_file=out_file,
xlabel='time',
ylabel='',
legend_entries=legend_entries,
outside=True)
out_file = os.path.join(
out_dir_viz, vid_file[:vid_file.rindex('.')] + '_' +
str(graph_num) + '.jpg')
visualize.saveMatAsImage(affinity, out_file)
visualize.writeHTMLForFolder(out_dir_viz)
def save_neg_cooc_graphs(out_dir):
all_file = os.path.join(out_dir,'all_classes_mul.npy')
all_cooc = np.load(all_file)
for class_name in class_names:
in_file = os.path.join(out_dir,class_name+'.npy')
curr_cooc = np.load(in_file)
out_cooc = curr_cooc - all_cooc
out_cooc = out_cooc + np.eye(out_cooc.shape[0])
out_file = os.path.join(out_dir,class_name+'neg.jpg')
visualize.saveMatAsImage(out_cooc, out_file)
# print 'curr_cooc',curr_cooc.shape,np.min(curr_cooc),np.max(curr_cooc)
# print 'out_cooc',out_cooc.shape,np.min(out_cooc),np.max(out_cooc)
# print 'all_cooc',all_cooc.shape,np.min(all_cooc),np.max(all_cooc)
# print out_file
out_file = os.path.join(out_dir,class_name+'neg.npy')
np.save(out_file, out_cooc)
visualize.writeHTMLForFolder(out_dir)
def save_neg_exp_cooc_graphs(out_dir):
for class_name in class_names:
in_file = os.path.join(out_dir,class_name+'neg.npy')
curr_cooc = np.load(in_file)
print np.min(curr_cooc),np.max(curr_cooc)
out_cooc = np.exp(curr_cooc-1)
print np.min(out_cooc),np.max(out_cooc)
out_file = os.path.join(out_dir,class_name+'negexp.jpg')
visualize.saveMatAsImage(out_cooc, out_file)
# print out_file
# print 'curr_cooc',curr_cooc.shape,np.min(curr_cooc),np.max(curr_cooc)
# print 'out_cooc',out_cooc.shape,np.min(out_cooc),np.max(out_cooc)
# print 'all_cooc',all_cooc.shape,np.min(all_cooc),np.max(all_cooc)
out_file = os.path.join(out_dir,class_name+'negexp.npy')
print out_file
np.save(out_file, out_cooc)
# raw_input()
visualize.writeHTMLForFolder(out_dir)
def save_sim_viz(vid_name,
out_shape_curr,
sim_mat,
class_idx,
out_dir,
dataset='ucf'):
gt_vals, det_times = get_gt_vector(vid_name,
out_shape_curr,
class_idx,
dataset=dataset)
if dataset.startswith('activitynet'):
class_names = globals.class_names_activitynet
else:
class_names = globals.class_names
out_dir_curr = os.path.join(out_dir, class_names[class_idx])
util.mkdir(out_dir_curr)
pos_rows = sim_mat[gt_vals > 0, :]
pos_rows = np.mean(pos_rows, axis=0)
neg_rows = sim_mat[gt_vals < 1, :]
neg_rows = np.mean(neg_rows, axis=0)
# for idx_pos_row, pos_row in enumerate(pos_rows):
max_val = max(np.max(pos_rows), np.max(neg_rows))
gt_vals_curr = gt_vals * max_val
arr_plot = [(det_times, curr_arr)
for curr_arr in [gt_vals_curr, pos_rows, neg_rows]]
legend_entries = ['gt', 'pos', 'neg']
# idx_pos_row = str(idx_pos_row)
out_file_curr = os.path.join(out_dir_curr, vid_name + '.jpg')
title = vid_name
# +' '+idx_pos_row
# visualize.plotSimple(arr_plot, out_file = out_file_curr, title = title, xlabel = 'time', ylabel = 'max sim', legend_entries = legend_entries)
# print out_file_curr
# print
np.save(out_file_curr.replace('.jpg', '.npy'), sim_mat)
visualize.saveMatAsImage(sim_mat, out_file_curr, title=title)
def get_distance_from_perfect(model_file, graph_num):
out_dir_meta = model_file[:model_file.rindex('.')]
out_dir_meta_meta = out_dir_meta + '_graph_etc'
out_dir_meta = out_dir_meta_meta + '_' + str(graph_num)
out_dir_viz = out_dir_meta + '_dist_perfectG'
print out_dir_viz
util.mkdir(out_dir_viz)
assert os.path.exists(out_dir_meta)
vid_files = glob.glob(os.path.join(out_dir_meta, '*validation*.npz'))
class_collations = [[] for idx in range(len(class_names))]
class_collations_pos = [[] for idx in range(len(class_names))]
viz = True
threshes = np.arange(0.1, 1.1, 0.1)
print threshes
for vid_file in vid_files:
print vid_file
npz_data = np.load(vid_file)
vid_file = os.path.split(vid_file)[1]
affinity = npz_data['affinity']
gt_vecs = npz_data['gt_vecs']
gt_classes = npz_data['gt_classes']
if viz:
out_file = os.path.join(
out_dir_viz, vid_file[:vid_file.rindex('.')] + '_' +
str(graph_num) + '.jpg')
visualize.saveMatAsImage(affinity, out_file)
plotter = []
legend_entries = []
for gt_idx, gt_class in enumerate(gt_classes):
gt_vec = gt_vecs[gt_idx]
gt_vec = gt_vec / np.max(gt_vec)
gt_vec = gt_vec * (gt_idx + 1)
x_axis = range(gt_vec.size)
plotter.append((x_axis, gt_vec))
legend_entries.append(class_names[gt_class])
out_file = os.path.join(
out_dir_viz, vid_file[:vid_file.rindex('.')] + '_gt.jpg')
visualize.plotSimple(plotter,
out_file=out_file,
xlabel='time',
ylabel='',
legend_entries=legend_entries)
for idx_gt, gt_vec in enumerate(gt_vecs):
gt_class = gt_classes[idx_gt]
class_name = class_names[gt_class]
gt_vec = gt_vec[:, np.newaxis]
perfectG = np.dot(gt_vec, gt_vec.T)
aff = np.array(affinity)
aff_just_pos = aff * perfectG
diff = get_l2_diff(aff, perfectG, threshes)
diff_pos = get_l2_diff(aff_just_pos, perfectG, threshes)
class_collations[gt_class].append(diff)
class_collations_pos[gt_class].append(diff_pos)
plotter = [(threshes, diff), (threshes, diff_pos)]
legend_entries = ['All', 'Pos']
out_file = os.path.join(
out_dir_viz, vid_file[:vid_file.rindex('.')] + '_' +
class_name + '_diff.jpg')
visualize.plotSimple(plotter,
out_file=out_file,
xlabel='Thresh',
ylabel='Diff',
legend_entries=legend_entries)
if viz:
out_file = os.path.join(
out_dir_viz, vid_file[:vid_file.rindex('.')] + '_' +
class_name + '_perfectG.jpg')
visualize.saveMatAsImage(perfectG, out_file)
visualize.writeHTMLForFolder(out_dir_viz)
for idx_class in range(len(class_names)):
class_name = class_names[idx_class]
cc = np.array(class_collations[idx_class])
ccp = np.array(class_collations_pos[idx_class])
cc = np.mean(cc, axis=0)
ccp = np.mean(ccp, axis=0)
plotter = [(threshes, cc), (threshes, ccp)]
legend_entries = ['All', 'Pos']
out_file = os.path.join(out_dir_viz,
'average_' + class_name + '_diff.jpg')
visualize.plotSimple(plotter,
out_file=out_file,
title=class_name,
xlabel='Thresh',
ylabel='Diff',
legend_entries=legend_entries)
visualize.writeHTMLForFolder(out_dir_viz)
def check_graph():
# model_file = '../experiments/graph_multi_video_pretrained_F_flexible_alt_temp_train_normalize_True_True_non_lin_HT_sparsify_True_num_switch_5_5_graph_size_32_focus_1_deno_8_n_classes_20_in_out_2048_64_2048_64_method_cos_pretrained_ucf_ucf/all_classes_False_just_primary_False_limit_500_cw_True_MultiCrossEntropy_500_step_500_0.1_0.0001_0.001_0.001_FIXED/model_199.pt'
model_file = '../experiments/graph_multi_video_pretrained_F_flexible_alt_train_temp_normalize_True_True_non_lin_HT_sparsify_True_num_switch_5_5_graph_size_32_focus_1_deno_8_n_classes_20_in_out_2048_64_2048_64_method_cos_pretrained_ucf_ucf/all_classes_False_just_primary_False_limit_500_cw_True_MultiCrossEntropy_500_step_500_0.1_0.0001_0.001_0.001_FIXED/model_299.pt'
model_file = '../experiments/graph_multi_video_pretrained_F_flexible_alt_train_temp_normalize_True_True_non_lin_HT_sparsify_True_num_switch_5_5_graph_size_2_focus_1_deno_8_n_classes_20_in_out_2048_64_2048_64_method_cos_pretrained_ucf_ucf/all_classes_False_just_primary_False_limit_500_cw_True_MultiCrossEntropy_500_step_500_0.1_0.0001_0.001_0.001_ABS/model_499.pt'
model_file = '../experiments/graph_multi_video_multi_F_joint_train_normalize_True_True_non_lin_HT_sparsify_True_graph_size_2_deno_8_n_classes_20_in_out_2048_64_feat_dim_2048_64_method_cos_ucf/all_classes_False_just_primary_False_limit_500_cw_True_MultiCrossEntropyMultiBranch_300_step_300_0.1_0.001_0.001_lw_0.5_0.5_ABS/model_299.pt'
model = torch.load(model_file).cuda()
model.eval()
train_data, test_train_data, test_data, n_classes, trim_preds = emb.get_data(
'ucf', 500, False, just_primary=False, gt_vec=False)
# test_data = train_data
# test_bool = False
# test_data.feature_limit = None
batch_size = 1
branch_to_test = 1
test_bool = True
out_dir_meta = model_file[:model_file.rindex('.')]
out_dir_meta = out_dir_meta + '_visualizing_' + str(branch_to_test)
util.mkdir(out_dir_meta)
print out_dir_meta
anno_file = test_data.anno_file
vid_names, annos = readTrainTestFile(anno_file)
test_dataloader = torch.utils.data.DataLoader(
test_data,
batch_size=batch_size,
collate_fn=test_data.collate_fn,
shuffle=False,
num_workers=1)
import torch.nn.functional as F
preds = []
labels = []
for idx_data, data in enumerate(test_dataloader):
gt_classes = np.where(annos[idx_data])[0]
vid_name = os.path.split(vid_names[idx_data])[1]
vid_name = vid_name[:vid_name.rindex('.')]
out_dir_curr = os.path.join(out_dir_meta, vid_name)
util.mkdir(out_dir_curr)
label = data['label'].cpu().data.numpy()
affinity = model.get_similarity(data['features'], sparsify=True)
x_all, pmf = model(data['features'], branch_to_test=branch_to_test)
assert len(pmf) == 1
x_all = torch.cat([x_all_curr.unsqueeze(0) for x_all_curr in x_all], 0)
x_all = F.softmax(x_all, dim=1)
x_all = x_all.data.cpu().numpy()
affinity = affinity.data.cpu().numpy()
for gt_class in gt_classes:
affinity_copy = np.array(affinity)
x_rel = x_all[:, gt_class]
thresh = np.max(x_rel) - (np.max(x_rel) - np.min(x_rel)) * 0.5
gt_vec, _ = get_gt_vector(vid_name,
x_rel.shape[0],
gt_class,
test=test_bool)
if np.sum(gt_vec) == 0:
'we got an anno problem', vid_name
continue
bin_keep = gt_vec.astype(int)
bin_keep_rot = np.roll(bin_keep, 1)
bin_keep_rot[0] = 0
diff = bin_keep - bin_keep_rot
idx_start_all = list(np.where(diff == 1)[0])
idx_end_all = list(np.where(diff == -1)[0])
idx_borders = np.array(idx_start_all + idx_end_all)
affinity_copy[:, idx_borders] = np.max(affinity_copy)
affinity_copy[idx_borders, :] = np.max(affinity_copy)
gt_vec = gt_vec * np.max(x_rel)
x_axis = range(x_rel.size)
thresh = thresh * np.ones(x_rel.shape)
out_file_curr = os.path.join(
out_dir_curr, 'det_confs_' + class_names[gt_class] + '.jpg')
visualize.plotSimple([(x_axis, x_rel), (x_axis, gt_vec),
(x_axis, thresh)],
out_file=out_file_curr,
title=class_names[gt_class],
xlabel='time',
ylabel='det conf',
legend_entries=['Det', 'GT', 'Thresh'])
out_file_mat = os.path.join(
out_dir_curr, 'mat_' + class_names[gt_class] + '.jpg')
visualize.saveMatAsImage(affinity_copy, out_file_mat)
preds.append(F.softmax(pmf[0]).data.cpu().numpy())
labels.append(label)
visualize.writeHTMLForFolder(out_dir_curr)
print out_dir_curr
raw_input()
labels = np.concatenate(labels, axis=0)
preds = np.concatenate(preds, axis=0)
labels[labels > 0] = 1
accuracy = sklearn.metrics.average_precision_score(labels, preds)
print accuracy
