app.py

import subprocess
import os  
if os.getenv('SYSTEM') == 'spaces':

    subprocess.call('pip install -U openmim'.split())
    subprocess.call('pip install python-dotenv'.split())
    subprocess.call('pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 torchaudio==0.12.1 --extra-index-url https://download.pytorch.org/whl/cu113'.split())
    subprocess.call('mim install mmcv>=2.0.0'.split())
    subprocess.call('mim install mmengine'.split())
    subprocess.call('mim install mmdet'.split())
    subprocess.call('pip install opencv-python-headless==4.5.5.64'.split())
    subprocess.call('pip install git+https://github.com/cocodataset/panopticapi.git'.split())

import gradio as gr

from huggingface_hub import snapshot_download
import cv2 
import dotenv 
dotenv.load_dotenv()
import numpy as np
import gradio as gr
import glob
from inference import inference_frame,inference_frame_serial
from inference import inference_frame_par_ready
from inference import process_frame
from inference import classes
from inference import class_sizes_lower
from metrics import process_results_for_plot
from metrics import prediction_dashboard
import os
import pathlib
import multiprocessing as mp
from time import time

if not os.path.exists('videos_example'):
    REPO_ID='SharkSpace/videos_examples'
    snapshot_download(repo_id=REPO_ID, token=os.environ.get('SHARK_MODEL'),repo_type='dataset',local_dir='videos_example')

theme = gr.themes.Soft(
    primary_hue="sky",
    neutral_hue="slate",
)



def add_border(frame, color = (255, 0, 0), thickness = 2):
    # Add a red border to the image
    relative = max(frame.shape[0],frame.shape[1])
    top = int(relative*0.025)
    bottom = int(relative*0.025)
    left = int(relative*0.025)
    right =  int(relative*0.025)
    # Add the border to the image
    bordered_image = cv2.copyMakeBorder(frame, top, bottom, left, right, cv2.BORDER_CONSTANT, value=color)
    
    return bordered_image 
    
def overlay_text_on_image(image, text_list, font=cv2.FONT_HERSHEY_SIMPLEX, font_size=0.5, font_thickness=1, margin=10, color=(255, 255, 255)):
    relative = min(image.shape[0],image.shape[1])
    y0, dy = margin, int(relative*0.1) # start y position and line gap
    for i, line in enumerate(text_list):
        y = y0 + i * dy
        text_width, _ = cv2.getTextSize(line, font, font_size, font_thickness)[0]
        cv2.putText(image, line, (image.shape[1] - text_width - margin, y), font, font_size, color, font_thickness, lineType=cv2.LINE_AA)
    return image

def draw_cockpit(frame, top_pred, cnt):
#     # Bullet points:
#     high_danger_color = (255,0,0)
#     low_danger_color = yellowgreen = (154,205,50)

#     shark_sighted = 'Shark Detected: ' + str(top_pred['shark_sighted'])
#     human_sighted = 'Number of Humans: ' + str(top_pred['human_n'])
#     shark_size_estimate = 'Biggest shark size: ' + str(top_pred['biggest_shark_size'])
#     shark_weight_estimate = 'Biggest shark weight: ' + str(top_pred['biggest_shark_weight'])
#     danger_level = 'Danger Level: ' 
#     danger_level += 'High' if top_pred['dangerous_dist'] else 'Low'
#     danger_color = high_danger_color if top_pred['dangerous_dist'] else low_danger_color # 'orangered' for high danger , 'yellowgreen' for low danger'
   
#    # Create a list of strings to plot
#     strings = [shark_sighted, human_sighted, shark_size_estimate, shark_weight_estimate, danger_level]
#     relative = max(frame.shape[0],frame.shape[1])
#     if top_pred['shark_sighted'] and (cnt % 2) == 0:
#         relative = max(frame.shape[0],frame.shape[1])
#         frame  = add_border(frame, color=danger_color, thickness=int(relative*0.025))
#     frame = overlay_text_on_image(frame, strings, font=cv2.FONT_HERSHEY_SIMPLEX, font_size=1.5, font_thickness=3, margin=int(relative*0.05), color=(255, 255, 255))
    return frame
    
def process_video(input_video, out_fps = 'auto', skip_frames = 7):
    cap = cv2.VideoCapture(input_video)

    output_path = "output.mp4"
    if out_fps != 'auto' and type(out_fps) == int:
        fps = int(out_fps)
    else:
        fps = int(cap.get(cv2.CAP_PROP_FPS))
        if out_fps == 'auto':
            fps = int(fps / skip_frames)

    width  = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
    height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))

    video = cv2.VideoWriter(output_path, cv2.VideoWriter_fourcc(*"mp4v"), fps, (width, height))

    iterating, frame = cap.read()
    cnt = 0
    
    while iterating:
        print(cnt)
        if (cnt % skip_frames) == 0:
            # Get inference results and output masks
            display_frame, result = inference_frame_serial(frame)
            video.write(cv2.cvtColor(display_frame, cv2.COLOR_BGR2RGB))
            
            #print(result)
            print('start top_pred')
            top_pred = process_results_for_plot(predictions = result.numpy(),
                                                classes = classes,
                                                class_sizes = class_sizes_lower)
            
            frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
            prediction_frame = cv2.cvtColor(display_frame, cv2.COLOR_BGR2RGB)
            
            frame = cv2.resize(frame, (int(width*4), int(height*4)))
            
            # Only show the prediction frame every other frame
            # print('first if')
            # if ((cnt*skip_frames) % 2 == 0): # and top_pred['shark_sighted']: 
            #     prediction_frame = cv2.resize(prediction_frame, (int(width*4), int(height*4)))
            #     frame = prediction_frame
             
            # # Add cockpit to frame
            # print('cockput if')
            # if  top_pred['shark_sighted']: 
            #     frame = draw_cockpit(frame, top_pred, cnt*skip_frames)
            
            print('dashboard')
            pred_dashbord = prediction_dashboard(top_pred = top_pred)
            #print('sending frame')
            print('finished dashboard')
            print(cnt)
            # if cnt == 119:
            #     import ipdb; ipdb.set_trace()
            
            yield prediction_frame, frame , None, pred_dashbord
        cnt += 1
        iterating, frame = cap.read()
        print('interating: ', iterating)
    
    video.release()
    yield None, None, output_path, None

with gr.Blocks(theme=theme) as demo:
    with gr.Row().style(equal_height=True,height='50%'):
        input_video = gr.Video(label="Input")
        output_video = gr.Video(label="Output Video")
        
    with gr.Row():
        processed_frames = gr.Image(label="Shark Engine")
        dashboard = gr.Image(label="Dashboard")
        original_frames = gr.Image(label="Original Frame") #, width='100%', height='100%')
    
    #with gr.Row(height='100%',width='100%'):
        #original_frames = gr.Image(label="Original Frame", width='100%', height='100%')
        
    with gr.Row():
        paths = sorted(pathlib.Path('videos_example/').rglob('*.mp4'))
        samples=[[path.as_posix()] for path in paths if 'raw_videos'  in str(path)]
        examples = gr.Examples(samples, inputs=input_video)
        process_video_btn = gr.Button("Process Video")

    process_video_btn.click(process_video, input_video, [processed_frames, original_frames, output_video, dashboard])

demo.queue()
if os.getenv('SYSTEM') == 'spaces':
    demo.launch(width='40%',auth=(os.environ.get('SHARK_USERNAME'), os.environ.get('SHARK_PASSWORD')), debug = True)
else: 
    demo.launch(debug = True)