This is the link to the dataset: https://data.caltech.edu/records/mzrjq-6wc02.

The dataset has images and respective contours in .mat files.
How to plot the contour in Python?

They have code that shows plotting the contour but it’s written in matlab:

function show_annotation(imgfile, annotation_file);
%% 
%% imgfile: string
%% annotation_file: string
%%
%% written by by Fei-Fei Li - November 2004
%%

IMTYPE = 'jpg'; 
GUIDELINE_MODE = 1; 
%% Parameters 
%label_abbrev = {'LE', 'RE', 'LN', 'NB', 'RN', 'LM', 'RM'}; 
LARGEFONT = 28; 
MEDFONT = 18; 
BIG_WINDOW = get(0,'ScreenSize'); 
SMALL_WINDOW = [100 100 512 480]; 
 
%% load the annotated data
load(annotation_file, 'box_coord', 'obj_contour');

%% Read and display image 
ima = imread(imgfile); 
ff=figure(1); clf; imagesc(ima); axis image; axis ij; hold on;
% black and white images
if length(size(ima))<3
   colormap(gray);
end
set(ff,'Position',SMALL_WINDOW); 
   
%% show box
box_handle = rectangle('position', [box_coord(3), box_coord(1), box_coord(4)-box_coord(3), box_coord(2)-box_coord(1)]);
set(box_handle, 'edgecolor','y', 'linewidth',5);

%% show contour
for cc = 1:size(obj_contour,2)
   if cc < size(obj_contour,2)
      plot([obj_contour(1,cc), obj_contour(1,cc+1)]+box_coord(3), [obj_contour(2,cc), obj_contour(2,cc+1)]+box_coord(1), 'r','linewidth',4);
   else
      plot([obj_contour(1,cc), obj_contour(1,1)]+box_coord(3), [obj_contour(2,cc), obj_contour(2,1)]+box_coord(1), 'r','linewidth',4);
   end
end

title(imgfile);

I do not know matlab but I was able to follow the provided script to plot a bounding box, but plotting the contour fails. This is my Python script:

img_dir = '/content/data/caltech101/101_ObjectCategories/chair/image_0037.jpg'
annot_dir = '/content/data/caltech101/Annotations/chair/annotation_0037.mat'

img = Image.open(img_dir_list[idx])
annot = sio.loadmat(annot_dir_list[idx])

obj_contour = annot['obj_contour']
print(obj_contour.shape[1])
for cc in range(obj_contour.shape[1]):
  if cc < obj_contour.shape[1] - 1:
    plt.plot([obj_contour[0,cc], obj_contour[0,cc+1]]+box_coord[3], [obj_contour[1,cc], obj_contour[1,cc+1]]+box_coord[1], 'r','linewidth',4)
  else:
    plt.plot([obj_contour[0,cc], obj_contour[0,0]]+box_coord[3], [obj_contour[1,cc], obj_contour[1,0]]+box_coord[1], 'r','linewidth',4)

I have added original image and contour map that the Python code produced here. As you can see the contour is flipped upside down.
If I subtract 200 from the y coordinate, the result looks ugly.