real hard

Dockerfile

@@ -1,12 +1,3 @@
-FROM node:21.2-alpine3.17 as build-step
-WORKDIR /yolov-app
-ENV PATH /app/node_modules/.bin:$PATH
-COPY package.json package-lock.json ./
-COPY ./src ./src
-COPY ./public ./public
-RUN yarn install
-RUN yarn build
-
 FROM python:3.10-bullseye
 
 RUN apt-get update && apt-get install -y \
@@ -29,7 +20,6 @@ ENV HOME=/home/user \
 	PATH=/home/user/.local/bin:$PATH
 
 WORKDIR $HOME/yolov-app
-COPY --chown=user --from=build-step /yolov-app/build ./build
 COPY --chown=user . $HOME/yolov-app
 
 
@@ -40,17 +30,17 @@ RUN gdown https://drive.google.com/file/d/1lBlZEh0v-WVUJsBb3ZJcUG8zxe5cckWT/view
 RUN gdown https://drive.google.com/file/d/14T5lpdJ0dPndYkMKr0sUcSq_myQDQSda/view?usp=sharing --fuzzy -O checkpoints/best181-8376/
 #ckpt_latest.pth
 
-RUN gdown https://drive.google.com/file/d/1pIDLMr0UUj_6fWYf9xHJ1lcKcCgmO7Vv/view?usp=sharing --fuzzy -O build/static/
+RUN gdown https://drive.google.com/file/d/1pIDLMr0UUj_6fWYf9xHJ1lcKcCgmO7Vv/view?usp=sharing --fuzzy -O uploads/
 #kitchen.mp4 
-RUN gdown https://drive.google.com/file/d/1ip07Lngcicd7pouKs_b0OhpVVoVFRC7U/view?usp=sharing --fuzzy -O build/static/
+RUN gdown https://drive.google.com/file/d/1ip07Lngcicd7pouKs_b0OhpVVoVFRC7U/view?usp=sharing --fuzzy -O uploads/
 #slip.mp4 
-RUN gdown https://drive.google.com/file/d/1s93CZeEKEp_SwO5Y4KOjNYK0zWBgNiy7/view?usp=sharing --fuzzy -O build/static/
+RUN gdown https://drive.google.com/file/d/1s93CZeEKEp_SwO5Y4KOjNYK0zWBgNiy7/view?usp=sharing --fuzzy -O uploads/
 #studycam.mp4 
-RUN gdown https://drive.google.com/file/d/1lZNim3Sl_dQN6gcyEvddRLeijM6v-pO5/view?usp=sharing --fuzzy -O build/static/
+RUN gdown https://drive.google.com/file/d/1lZNim3Sl_dQN6gcyEvddRLeijM6v-pO5/view?usp=sharing --fuzzy -O uploads/
 #cafe_fall.mp4 
-RUN gdown https://drive.google.com/file/d/1Ru6ARgQOtshVgSUosFiH7GFRyRM-_R5U/view?usp=sharing --fuzzy -O build/static/
+RUN gdown https://drive.google.com/file/d/1Ru6ARgQOtshVgSUosFiH7GFRyRM-_R5U/view?usp=sharing --fuzzy -O uploads/
 #skate.mp4
-RUN gdown https://drive.google.com/file/d/1Md80GemqcuWhVDKXftNNt6owCQQQlWxi/view?usp=sharing --fuzzy -O build/static/
+RUN gdown https://drive.google.com/file/d/1Md80GemqcuWhVDKXftNNt6owCQQQlWxi/view?usp=sharing --fuzzy -O uploads/
 #Sample.png
 
 EXPOSE 5000

app.py

@@ -0,0 +1,211 @@
+import gradio as gr
+from super_gradients.training import models
+from deep_sort_torch.deep_sort.deep_sort import DeepSort
+from super_gradients.training import models
+from super_gradients.training.pipelines.pipelines import DetectionPipeline
+from model_tools import  get_prediction, get_color
+import cv2
+import datetime
+import torch
+import os
+import gradio as gr
+import numpy as np
+
+np.float = float  
+np.int = int 
+np.object = object   
+np.bool = bool
+
+dir = os.getcwd()+ '/uploads/'
+
+inp = gr.Image(type="pil")
+output = gr.Image(type="pil")
+
+examples=[[dir +"cafe_fall.mp4","Fall in cafe"],
+          [dir +"slip.mp4","Run and Fall2"],
+          [dir +"skate.mp4","Skate and Fall"],
+          [dir +"kitchen.mp4","Fall in kitchen"],
+          [dir +"studycam.mp4","Experiment fall"]] 
+
+ckpt_path =  os.getcwd() + "/checkpoints/best181-8376/ckpt_latest.pth"
+best_model = models.get('yolo_nas_s',
+                        num_classes=1,
+                        checkpoint_path=ckpt_path)
+
+best_model = best_model.to("cuda" if torch.cuda.is_available() else "cpu")                        
+#best_model = models.get("yolo_nas_s", pretrained_weights="coco")                        
+best_model.eval()
+
+#### Initiatize tracker
+tracker_model = os.getcwd() + "/checkpoints/ckpt.t7"
+tracker = DeepSort(model_path=tracker_model,max_age=30,nn_budget=100, max_iou_distance=0.7, max_dist=0.2)
+out_path=dir 
+filename = 'demo.webm'
+
+description = "Yolo model to detect if a person is falling or fallen with deepsort to track how long the subject has fallen.\
+            If the duration crosses a threshold of 5s, the bounding box will turn red and the subject be labelled as IMMOBILE."
+
+def vid_predict(media):
+
+    pipeline = DetectionPipeline(
+                model=best_model,
+                image_processor=best_model._image_processor,
+                post_prediction_callback=best_model.get_post_prediction_callback(iou=0.25, conf=0.70),
+                class_names=best_model._class_names,
+            )    
+    
+    print("Running Predict")
+    save_to = os.path.join(out_path, filename)
+    cap = cv2.VideoCapture(media)
+    
+    if cap.isOpened():
+
+        width  = cap.get(3)  # float `widtqh`
+        print('width',width)
+        height = cap.get(4)
+        print('Height',height)
+        fps = cap.get(cv2.CAP_PROP_FPS)
+        # or
+        fps = cap.get(5)
+        
+        print('fps:', fps)  # float `fps`
+        
+        frame_count = cap.get(cv2.CAP_PROP_FRAME_COUNT)
+        # or frame_count = cap.get(7)
+        
+        print('frames count:', frame_count)  # float `frame_count`    
+     
+    out = cv2.VideoWriter(save_to, cv2.VideoWriter_fourcc(*'VP08'), fps, (640,640))
+    fall_records = {}
+    frame_id = 0
+    while True:
+        frame_id += 1
+        if frame_id > frame_count:
+                break
+        print('frame_id', frame_id)
+    
+        ret, img = cap.read()
+        #img = cv2.resize(img, (1280, 720),cv2.INTER_AREA)
+        # if height > 720:
+        #     print("Reshaped")
+        img = cv2.resize(img, (640, 640),cv2.INTER_AREA)
+        width, height =  img.shape[1],  img.shape[0]
+
+        ### recalibrate color channels to rgb for use in model prediction 
+        img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
+        overlay = img.copy()
+
+        ### create list objects needed for tracking
+        detects = []
+        conffs = []
+
+
+        if ret:
+            print("START ")    
+            model_predictions = get_prediction(best_model, img_rgb, pipeline)
+            print(model_predictions)
+            classnames = ['Fall-Detected']
+            results = model_predictions
+            bboxes = results.bboxes_xyxy
+            
+            if len(bboxes) >= 1:
+                confs = results.confidence
+                labels = results.labels
+
+                for bbox, conf, label in zip(bboxes, confs, labels):
+                    label = int(label)
+                    conf = np.round(conf, decimals=2)
+
+                    x1, y1, x2, y2 = bbox[0], bbox[1], bbox[2], bbox[3]
+                    x1, y1, x2, y2 = int(x1), int(y1), int(x2), int(y2)
+
+                    ### for tracking model
+                    bw = abs(x1 - x2)
+                    bh = abs(y1 - y2)
+                    cx , cy = x1 + bw//2, y1 + bh//2
+
+                    coords = [cx, cy, bw, bh]
+                    detects.append(coords)
+                    conffs.append([float(conf)])     
+           
+                ### Tracker
+                xywhs = torch.tensor(detects)
+                conffs = torch.tensor(conffs)
+                #tracker_results = deepsort.update(xywhs, confss,oids, img)
+                tracker_results = tracker.update(xywhs, conffs, img_rgb)
+
+                ### conduct check on track_records
+                now = datetime.datetime.now()
+                if len(fall_records.keys()) >=1:
+                    #print(fall_records)
+
+                    ### reset timer for calculating immobility to 0 if time lapsed since last detection of fall more than N seconds
+                    fall_records = {id: item if (now - item['present']).total_seconds() <= 3.0 else  {'start':now, 'present': now} for id, item in fall_records.items() }
+
+
+                if len(tracker_results)>=1:            
+                    for track,conf,label in zip(tracker_results,conffs, labels):
+                        conf = conf.numpy()[0]
+                        duration = 0
+                        minute = 0
+                        sec = 0
+                        x1, y1 ,x2, y2, id  = track
+                        x1, y1, x2, y2 = int(x1), int(y1), int(x2), int(y2)
+
+                        if id in fall_records.keys():
+                            ### record present time
+                            present = datetime.datetime.now()
+                            fall_records[id].update({'present': present})
+
+                            ### calculate duration 
+                            duration =  fall_records[id]['present'] - fall_records[id]['start']
+                            duration = int(duration.total_seconds())
+                            
+                            ### record status
+                            fall_records[id].update({'status': 'IMMOBILE'}) if duration >= 5 else fall_records[id].update({'status': None})
+                            print(f"Frame:{frame_id} ID:  {id} Conf: {conf}  Duration:{duration} Status: {fall_records[id]['status']}")
+                            print(fall_records[id])
+                            minute, sec = divmod(duration,60)
+                            
+                        else:
+                            start = datetime.datetime.now()
+                            fall_records[id] = {'start': start}
+                            fall_records[id].update({'present': start})
+
+                        classname = classnames[int(label)]
+    
+
+                        color = get_color(id*20)
+                        if duration < 5:
+                            display_text = f"{str(classname)} ({str(id)}) {str(conf)} Elapsed: {round(minute)}min{round(sec)}s"
+                            (w, h), _ = cv2.getTextSize(
+                                display_text, cv2.FONT_HERSHEY_SIMPLEX, 0.7, 1)  
+                            cv2.rectangle(img,(x1, y1), (x2, y2),color,1)
+                            cv2.rectangle(overlay,(x1, y1), (x2, y2),color,1)
+                            cv2.rectangle(overlay, (min(x1,int(width)-w), max(1,y1 - 20)), (min(x1+ w,int(width)) , max(21,y1)), color, cv2.FILLED) 
+                        else:
+                            display_text = f"{str(classname)} ({str(id)}) {str(conf)} IMMOBILE: {round(minute)}min{round(sec)}s "
+                            (w, h), _ = cv2.getTextSize(
+                                display_text, cv2.FONT_HERSHEY_SIMPLEX, 0.7, 1)                        
+                            cv2.rectangle(img,(x1, y1), (x2, y2),(0,0,255),1)
+                            cv2.rectangle(overlay,(x1, y1), (x2, y2),(0,0,255),1)
+                            cv2.rectangle(overlay, (min(x1,int(width)-w), max(1,y1 - 20)), (min(x1+ w,int(width)) , max(21,y1)), (0,0,255), cv2.FILLED) 
+
+                        cv2.putText(img,display_text, (min(x1,int(width)-w), max(21,y1)), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0,0,0),2)
+                        cv2.putText(overlay,display_text, (min(x1,int(width)-w), max(21,y1)), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0,0,0),2)                    
+                        
+        alpha = 0.6
+        masked = cv2.addWeighted(overlay, alpha, img, 1 - alpha, 0)
+        out.write(masked)
+ 
+    cap.release()
+    out.release() 
+
+    cv2.destroyAllWindows()
+
+    return save_to
+
+demo = gr.Interface(fn=vid_predict, inputs=gr.Video(), outputs=gr.Video(), examples=examples, description=description, title='Fall detection and tracking with deep sort')
+    
+if __name__ == "__main__":
+    demo.launch(show_api=False) 

main.py

@@ -1,114 +0,0 @@
-from flask import Flask, request, render_template, send_from_directory
-from flask import flash, request, redirect, url_for, Response, make_response
-from werkzeug.utils import secure_filename
-
-from super_gradients.training import models
-from deep_sort_torch.deep_sort.deep_sort import DeepSort
-
-import torch
-
-from model_tools import vid_predict, img_predict
-from dotenv import load_dotenv
-import os
-import urllib.request
-
-load_dotenv()
-secret_key = os.getenv("secret_key")
-
-dir = os.getcwd()+ f'/build'
-dir_static= dir + '/static'
-dir_ckpt = os.getcwd()+ f'/checkpoints'
-
-ckpt_path =  dir_ckpt + "/best181-8376/ckpt_latest.pth"
-best_model = models.get('yolo_nas_s',
-                        num_classes=1,
-                        checkpoint_path=ckpt_path)
-
-best_model = best_model.to("cuda" if torch.cuda.is_available() else "cpu")                                             
-best_model.eval()
-
-#### Initiatize tracker
-tracker_model = "./checkpoints/ckpt.t7"
-tracker = DeepSort(model_path=tracker_model,max_age=30,nn_budget=100, max_iou_distance=0.7, max_dist=0.2)
-
-
-ALLOWED_EXTENSIONS = {'png', 'jpg', 'jpeg', 'gif', 'mp4', 'avi','webm'}
-
-app = Flask(__name__, template_folder=dir,static_folder=dir_static)
-app.config['UPLOAD_FOLDER'] = dir_static
-app.config['MAX_CONTENT_LENGTH'] = 20*1024*1024
-app.secret_key = secret_key
-
-def allowed_file(filename):
-    return '.' in filename and \
-           filename.rsplit('.', 1)[1].lower() in ALLOWED_EXTENSIONS
-
-@app.route('/')
-def index():
-    predictions = False
-    return render_template('index.html', predictions=predictions)
-
-@app.route('/upload', methods=["GET", "POST"])
-def upload():
-
-    if request.method == 'POST':
-        print('Form',request.form.get('options'))
-        try:
-            filename = request.form.get('options')
-            if filename:
-                save_path = os.path.join(app.config['UPLOAD_FOLDER'], filename)
-                filetype = 'video'
-                new_filename = filename[:-4] + ".webm"
-                save_to = vid_predict(save_path,best_model,tracker, out_path=dir_static, filename=new_filename)
-                save_to = url_for('static', filename=new_filename)
-                predictions = True
-                return render_template('index.html', predictions=predictions, saved_outout=save_to, ft=filetype)
-        except:
-            pass    
-        # check if the post request has the file part
-        if ('file') and ('media') not in request.files:
-            flash('No file part')
-            return redirect(request.url)
-        
-        try:
-            file = request.files['file']
-        except:
-            file = request.files['media']
-        # If the user does not select a file, the browser submits an
-        # empty file without a filename.
-        if file.filename == '':
-            flash('No selected file')
-            return redirect(request.url)
-        
-        if file and allowed_file(file.filename):
-            filename = secure_filename(file.filename)
-            save_path = os.path.join(app.config['UPLOAD_FOLDER'], filename)
-            file.save(save_path)
-            new_filename = filename[:-4] + ".webm"
-
-            if filename[-3:] in ['mp4','avi','webm']:
-                print("VIDEO")
-                filetype = 'video'
-                save_to = vid_predict(save_path,best_model,tracker, out_path=dir_static, filename=new_filename)
-                save_to = url_for('static', filename=new_filename)
-            else:
-                print("IMAGE")
-                filetype = 'image'
-                save_to = img_predict(save_path,best_model, out_path=dir_static, filename=new_filename)
-                save_to = url_for('static', filename="pred_0.jpg")
-
-            predictions = True
-           
-            return render_template('index.html', predictions=predictions, saved_outout=save_to, ft=filetype)
-
-
-
-@app.route('/static/<folder>/<file>')
-def css(folder,file):
-    ''' User will call with with thier id to store the symbol as registered'''   
-    path = folder+'/'+file
-    return send_from_directory(directory=dir_static,path=path)
-
-                                
-if __name__ == "__main__":
-    app.run(host="0.0.0.0", port=7860)

model_tools.py

@@ -4,18 +4,22 @@ from matplotlib.colors import hsv_to_rgb
 import torch
 import numpy as np
 from super_gradients.training import models
+from super_gradients.training.models.detection_models.customizable_detector import CustomizableDetector
+from super_gradients.training.pipelines.pipelines import DetectionPipeline
+
 from deep_sort_torch.deep_sort.deep_sort import DeepSort
 import os
 
+# make sure to set IOU and confidence in the pipeline constructor
 
 def get_color(number):
     """ Converts an integer number to a color """
-
+    # change these however you want to
     hue = number*30 % 180
     saturation = number*103 % 256
     value = number*50 % 256
 
-
+    # expects normalized values
     hsv_array = [hue/179, saturation/255, value/255]
     rgb = hsv_to_rgb(hsv_array)
 
@@ -28,179 +32,25 @@ def img_predict(media, model, out_path,filename):
 
     return None
 
+def get_prediction(model, image_in, pipeline):
+    ''' Obtains DetectionPrediction object from a single input RGB image
+    '''
+    # Preprocess
+    preprocessed_image, processing_metadata = pipeline.image_processor.preprocess_image(image=image_in.copy())
+
+    # Predict
+    with torch.no_grad():
+        torch_input = torch.Tensor(preprocessed_image).unsqueeze(0).to('cuda')
+        model_output = model(torch_input)
+        prediction = pipeline._decode_model_output(model_output, model_input=torch_input)
+  # Postprocess
+    return pipeline.image_processor.postprocess_predictions(predictions=prediction[0], metadata=processing_metadata)
+
+
 
-def vid_predict(media, model, tracker, out_path,filename):
-    print("Running Predict")
-    save_to = os.path.join(out_path, filename)
-    cap = cv2.VideoCapture(media)
-    
-    if cap.isOpened():
-
-        width  = cap.get(3)  # float `widtqh`
-        print('width',width)
-        height = cap.get(4)
-        print('Height',height)
-        fps = cap.get(cv2.CAP_PROP_FPS)
-        # or
-        fps = cap.get(5)
-        
-        print('fps:', fps)  # float `fps`
-        
-        frame_count = cap.get(cv2.CAP_PROP_FRAME_COUNT)
-        # or frame_count = cap.get(7)
-        
-        print('frames count:', frame_count)  # float `frame_count`    
-     
-    out = cv2.VideoWriter(save_to, cv2.VideoWriter_fourcc(*'VP08'), fps, (640,640))
-    fall_records = {}
-    frame_id = 0
-    while True:
-        frame_id += 1
-        if frame_id > frame_count:
-                break
-        print('frame_id', frame_id)
-    
-        ret, img = cap.read()
-        img = cv2.resize(img, (640, 640),cv2.INTER_AREA)
-        width, height =  img.shape[1],  img.shape[0]
-
-        ### recalibrate color channels to rgb for use in model prediction 
-        img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
-        overlay = img.copy()
-
-        ### create list objects needed for tracking
-        detects = []
-        conffs = []
-
-
-        if ret:
-
-            model_predictions = model.predict(img_rgb,conf=0.70,fuse_model=False)
-            classnames = model_predictions[0].class_names
-            results = model_predictions[0].prediction
-            bboxes = results.bboxes_xyxy
-            
-            if len(bboxes) >= 1:
-                confs = results.confidence
-                labels = results.labels
-
-                for bbox, conf, label in zip(bboxes, confs, labels):
-                    label = int(label)
-                    conf = np.round(conf, decimals=2)
-  
-                
-
-                    x1, y1, x2, y2 = bbox[0], bbox[1], bbox[2], bbox[3]
-                    x1, y1, x2, y2 = int(x1), int(y1), int(x2), int(y2)
-
-                    ### for tracking model
-                    bw = abs(x1 - x2)
-                    bh = abs(y1 - y2)
-                    cx , cy = x1 + bw//2, y1 + bh//2
-
-                    coords = [cx, cy, bw, bh]
-                    detects.append(coords)
-                    conffs.append([float(conf)])     
-           
-                ### Tracker
-                xywhs = torch.tensor(detects)
-                conffs = torch.tensor(conffs)
-                tracker_results = tracker.update(xywhs, conffs, img_rgb)
-
-                ### conduct check on track_records
-                now = datetime.datetime.now()
-                if len(fall_records.keys()) >=1:
-         
-
-                    ### reset timer for calculating immobility to 0 if time lapsed since last detection of fall more than N seconds
-                    fall_records = {id: item if (now - item['present']).total_seconds() <= 3.0 else  {'start':now, 'present': now} for id, item in fall_records.items() }
-
-
-                if len(tracker_results)>=1:            
-                    for track,conf,label in zip(tracker_results,conffs, labels):
-                        conf = conf.numpy()[0]
-                        duration = 0
-                        minute = 0
-                        sec = 0
-                        x1, y1 ,x2, y2, id  = track
-                        x1, y1, x2, y2 = int(x1), int(y1), int(x2), int(y2)
-
-                        if id in fall_records.keys():
-                            ### record present time
-                            present = datetime.datetime.now()
-                            fall_records[id].update({'present': present})
-
-                            ### calculate duration 
-                            duration =  fall_records[id]['present'] - fall_records[id]['start']
-                            duration = int(duration.total_seconds())
-                            
-                            ### record status
-                            fall_records[id].update({'status': 'IMMOBILE'}) if duration >= 5 else fall_records[id].update({'status': None})
-                            print(f"Frame:{frame_id} ID:  {id} Conf: {conf}  Duration:{duration} Status: {fall_records[id]['status']}")
-                            print(fall_records[id])
-                            minute, sec = divmod(duration,60)
-                            
-                        else:
-                            start = datetime.datetime.now()
-                            fall_records[id] = {'start': start}
-                            fall_records[id].update({'present': start})
-
-                        classname = classnames[int(label)]
-    
 
-                        color = get_color(id*20)
-                        if duration < 5:
-                            display_text = f"{str(classname)} ({str(id)}) {str(conf)} Elapsed: {round(minute)}min{round(sec)}s"
-                            (w, h), _ = cv2.getTextSize(
-                                display_text, cv2.FONT_HERSHEY_SIMPLEX, 0.7, 1)  
-                            cv2.rectangle(img,(x1, y1), (x2, y2),color,1)
-                            cv2.rectangle(overlay,(x1, y1), (x2, y2),color,1)
-                            cv2.rectangle(overlay, (min(x1,int(width)-w), max(1,y1 - 20)), (min(x1+ w,int(width)) , max(21,y1)), color, cv2.FILLED) 
-                        else:
-                            display_text = f"{str(classname)} ({str(id)}) {str(conf)} IMMOBILE: {round(minute)}min{round(sec)}s "
-                            (w, h), _ = cv2.getTextSize(
-                                display_text, cv2.FONT_HERSHEY_SIMPLEX, 0.7, 1)                        
-                            cv2.rectangle(img,(x1, y1), (x2, y2),(0,0,255),1)
-                            cv2.rectangle(overlay,(x1, y1), (x2, y2),(0,0,255),1)
-                            cv2.rectangle(overlay, (min(x1,int(width)-w), max(1,y1 - 20)), (min(x1+ w,int(width)) , max(21,y1)), (0,0,255), cv2.FILLED) 
-
-                        cv2.putText(img,display_text, (min(x1,int(width)-w), max(21,y1)), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0,0,0),2)
-                        cv2.putText(overlay,display_text, (min(x1,int(width)-w), max(21,y1)), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0,0,0),2)                    
-                        
-      
-                
-            ### output image
-        alpha = 0.6
-        masked = cv2.addWeighted(overlay, alpha, img, 1 - alpha, 0)
-        out.write(masked)
-     
-
-    cap.release()
-    out.release() 
-
-    cv2.destroyAllWindows()
-
-
-
-if __name__ == '__main__':
-    #ckpt_path = "/home/kaelan/Projects/Jupyter/Pytorch/Yolo-Nas/yolov-app/checkpoints/ckpt_latest.pth"
-    ckpt_path = "/home/kaelan/Projects/Jupyter/Pytorch/Yolo-Nas/checkpoints_Fall_detection/Fall_yolonas_run2/ckpt_latest.pth"
     
-    best_model = models.get('yolo_nas_s',
-                            num_classes=1,
-                            checkpoint_path=ckpt_path)
-    # best_model.set_dataset_processing_params(
-    # class_names=['Fall-Detected'],
-    # iou=0.35, conf=0.7,
-    # )
-    best_model = best_model.to("cuda" if torch.cuda.is_available() else "cpu")                        
-    #best_model = models.get("yolo_nas_s", pretrained_weights="coco")                        
-    best_model.eval()
-
-    #### Initiatize tracker
-    tracker_model = "./checkpoints/ckpt.t7"
-    tracker = DeepSort(model_path=tracker_model,max_age=30,nn_budget=100, max_iou_distance=0.7, max_dist=0.2)    
-
-    title = "skate.mp4"
-    media = "/home/kaelan/Projects/data/videos/" + title
-    vid_predict(media,best_model,tracker) 
+
+
+
+

package.json

@@ -1,40 +0,0 @@
-{
-  "name": "yolov-app",
-  "version": "0.1.0",
-  "private": true,
-  "dependencies": {
-    "axios": "^1.6.2",
-    "react": "^18.2.0",
-    "react-dom": "^18.2.0",
-    "react-scripts": "5.0.1"
-  },
-  "scripts": {
-    "start": "react-scripts start",
-    "build": "react-scripts build",
-    "test": "react-scripts test",
-    "eject": "react-scripts eject"
-  },
-  "eslintConfig": {
-    "extends": [
-      "react-app",
-      "react-app/jest"
-    ]
-  },
-  "browserslist": {
-    "production": [
-      ">0.2%",
-      "not dead",
-      "not op_mini all"
-    ],
-    "development": [
-      "last 1 chrome version",
-      "last 1 firefox version",
-      "last 1 safari version"
-    ]
-  },
-  "devDependencies": {
-    "dart-sass": "^1.25.0",
-    "node-sass": "npm:dart-sass@^1.25.0",
-    "node-sass-install": "^1.0.2"
-  }
-}

public/index.html

@@ -1,112 +0,0 @@
-<!DOCTYPE html>
-<html lang="en">
-
-<head>
-    <meta charset="utf-8" />
-    <link rel="icon" href="%PUBLIC_URL%/favicon.ico" />
-    <link rel="stylesheet" href="{{url_for('static', filename='css/index.css')}}">
-    <meta name="viewport" content="width=device-width, initial-scale=1" />
-    <meta name="theme-color" content="#000000" />
-    <meta name="description" content="Web site created using create-react-app" />
-    <link rel="apple-touch-icon" href="%PUBLIC_URL%/logo192.png" />
-    <!--
-      manifest.json provides metadata used when your web app is installed on a
-      user's mobile device or desktop. See https://developers.google.com/web/fundamentals/web-app-manifest/
-    -->
-    <link rel="manifest" href="%PUBLIC_URL%/manifest.json" />
-    <!--
-      Notice the use of %PUBLIC_URL% in the tags above.
-      It will be replaced with the URL of the `public` folder during the build.
-      Only files inside the `public` folder can be referenced from the HTML.
-
-      Unlike "/favicon.ico" or "favicon.ico", "%PUBLIC_URL%/favicon.ico" will
-      work correctly both with client-side routing and a non-root public URL.
-      Learn how to configure a non-root public URL by running `npm run build`.
-    -->
-    <link rel="preconnect" href="https://fonts.gstatic.com">
-    <link href="https://fonts.googleapis.com/css2?family=Poppins:wght@300;400;600&display=swap" rel="stylesheet">
-    <link href="https://cdn.jsdelivr.net/npm/bootstrap@5.3.2/dist/css/bootstrap.min.css" rel="stylesheet" integrity="sha384-T3c6CoIi6uLrA9TneNEoa7RxnatzjcDSCmG1MXxSR1GAsXEV/Dwwykc2MPK8M2HN" crossorigin="anonymous">
-    <title>FALL DETECTION App</title>
-
-  </head>
-
-
-<body>
-    <noscript>You need to enable JavaScript to run this app.</noscript>
-    <script src="https://cdn.jsdelivr.net/npm/bootstrap@5.3.2/dist/js/bootstrap.bundle.min.js" integrity="sha384-C6RzsynM9kWDrMNeT87bh95OGNyZPhcTNXj1NW7RuBCsyN/o0jlpcV8Qyq46cDfL" crossorigin="anonymous"></script>
-    <div id="content">  
-    <nav class="navbar navbar-dark bg-dark">
-      <a class="navbar-brand" style="color: #8B008B" href="#">FALL DETECTION</a>
-        
-    </nav>   
-
-   <!--OFF Canvas-->>
-   <br>
-   <div class="text-center">
-    <button class="btn btn-dark btn-lg" type="button" data-bs-toggle="offcanvas" data-bs-target="#offcanvasTop" aria-controls="offcanvasTop">Toggle Introduction to Experiment</button>
-  </div>
-  
-  
-
-   <div class="offcanvas offcanvas-top" tabindex="-1" id="offcanvasTop" aria-labelledby="offcanvasTopLabel" style="background-color: #0b0b0bcf">
-     <div class="offcanvas-header">
-       <h5 id="offcanvasTopLabel">Introduction on experiment</h5>
-
-       <button type="button" class="btn-close text-reset" data-bs-dismiss="offcanvas" aria-label="Close"></button>
-     </div>
-     <div class="offcanvas-body">
-      
-      <h2 style="color: #FEFBEA"> Introduction on experiment </h2>
-      <img src="{{ url_for('static', filename='Sample.png') }}" ,width="240" height="240">
-      <ol>
-         <li style="color: #FEFBEA"><b style="color: #FEFBEA"> Experimentation on using YOLO-NAS model(s-size) to do fall recognition. The model is trained on images of various scenarios of people falling or have fallen.</b></li>
-         <li style="color: #FEFBEA"><b style="color: #FEFBEA">Deepsort alogrithm is use to track falls. There will be an id number next to the classification label "FALL DETECTED" and confidence in prediction </b></li>
-         <li style="color: #FEFBEA"><b style="color: #FEFBEA">For this experiment if an individual has been tracked as fallen for more than 5 seconds, the bounding box of that personal will turn red and be tagged as "IMMOBILE".</b></li>
-         <li style="color: #FEFBEA"><b style="color: #FEFBEA">The model has achieve more than 0.8 score on @map50 which will serve as a good baseline </b>
-          <li style="color: #FEFBEA"><b style="color: #FEFBEA"> More datasets can be downloaded to test on the model at http://www.iro.umontreal.ca/~labimage/Dataset/ and https://www.kaggle.com/datasets/tuyenldvn/falldataset-imvia</b>
-     </ol>
-     </div>
-   </div>   
-    
-
-    <div class="container text-center">
-      <div class="row align-items-start">
-
-        <div class="col" id="root">
-        </div>
-
-        <div class="col">
-          <h1>RESULT</h1>
-          <p>Video will take afew minutes or more to process. Upload images for faster processing </p>  
-            {% if predictions and ft=='video': %} 
-            <video class="border border-secondary" controls autoplay><source src={{saved_outout}} type="video/mp4" height='680' width='680' name={{predictions}}></video>
-            {% elif predictions and ft=='image': %}
-            <img class="border border-secondary" src={{saved_outout}}  style="width:640px;height:640px;"> 
-            {% else %}
-            <div class="square"></div> 
-          {% endif %}
-        </div>
-
-      </div>
-    </div>
-
-
-    <!---<div id="root">
- 
-    </div>--->
- 
-    <!--
-      This HTML file is a template.
-      If you open it directly in the browser, you will see an empty page.
-
-      You can add webfonts, meta tags, or analytics to this file.
-      The build step will place the bundled scripts into the <body> tag.
-
-      To begin the development, run `npm start` or `yarn start`.
-      To create a production bundle, use `npm run build` or `yarn build`.
-   
-    --> 
-  </div>  
-</body>
-
-</html>

public/manifest.json

@@ -1,25 +0,0 @@
-{
-  "short_name": "React App",
-  "name": "Create React App Sample",
-  "icons": [
-    {
-      "src": "favicon.ico",
-      "sizes": "64x64 32x32 24x24 16x16",
-      "type": "image/x-icon"
-    },
-    {
-      "src": "logo192.png",
-      "type": "image/png",
-      "sizes": "192x192"
-    },
-    {
-      "src": "logo512.png",
-      "type": "image/png",
-      "sizes": "512x512"
-    }
-  ],
-  "start_url": ".",
-  "display": "standalone",
-  "theme_color": "#000000",
-  "background_color": "#ffffff"
-}

public/robots.txt

@@ -1,3 +0,0 @@
-# https://www.robotstxt.org/robotstxt.html
-User-agent: *
-Disallow:

requirements.txt

@@ -161,5 +161,6 @@ youtube-dl==2021.12.17
 uvicorn
 yt-dlp==2023.10.13
 gunicorn
+gradio
 gdown
 zipp

src/App.js

@@ -1,121 +0,0 @@
-import { useState,useEffect,useRef } from 'react';
-
-window.addEventListener("click",() => {
-  const loader = document.querySelector(".loader");
-
-  loader.classList.add("loader-hidden");
-
-  loader.addEventListener("transitionend",() => {
-    document.body.removeChild("loader")
-  })
-})
-   
-
-function App() {
-  const [imageURL, setImageURL] = useState(null);
-  const imageRef = useRef()
-  const [imageURL1, setImageURL1] = useState(null);
-  const [imageURL2, setImageURL2] = useState(null);
-  const [file_img, setfile_img] = useState(null);
-  const [file_video, setfile_video] = useState(null);
-  const imageRef2 = useRef()
-
-  const uploadImage = (e) => {
-    console.log(e)
-    const{ files } = e.target
-    const filetype = e.target.files[0].type.slice(0,5)
-    console.log(filetype)
-
-    if (filetype === 'image') {
-      setfile_img(filetype)
-      setfile_video(null)
-    } else if (filetype === 'video') {
-      setfile_video(filetype)
-      setfile_img(null)
-
-    } else {
-      setfile_img(null)
-      setfile_video(null)
-    }
-
-    if (files.length > 0) {
-      const url = URL.createObjectURL(files[0])
-      //const a = document.createElement('a');
-      //a.href = url
-      //console.log('A', a)
-      //a.download = 'save01.mp4';
-      //a.click();     
-
-      setImageURL(url)
-      setImageURL1(url)
-    } else {
-      setImageURL(null)
-      setImageURL1(null)
-    }
-  } 
-
-  const uploadName = (e) => {
-    console.log(e.target.value)
-    
-    const files = e.target.value
-    var upath =  "/static/"
-    const url = upath.concat(files)
-
-
-    setImageURL(url)
-    setImageURL2(url)
-    setfile_video(url)
-    setfile_img(null)
-
-
-  } 
-
-  return (
-    <div className="App">
-      <h1 className='header'>PREVIEW</h1>
-
-        <div className='inputHolder' >
-          <form action="/upload" enctype="multipart/form-data" method="post" id='uploadform'>
-            <div class="mb-3">
-
-              <label for="formFile" class="form-label">Upload Image or Video to preview:   Max(20mb/file)</label>
-              <a href="https://drive.google.com/drive/folders/1Gslj0uO2CeLXsO9EhjJ0OepWWpyWb6yO?usp=sharing" target="_blank">Sample Pics to download</a> 
-              <input class="form-control" type='file' id="formFile" accepts='image/*, video/*' capture='camera' name='media'
-                onChange={uploadImage}/>
-
-                {imageURL1 && <button type="submit" class="btn btn-dark" value='SubmiT' >DETECT FALL </button>}  
-            </div>
-
-              </form>
-              <br></br>
-             <div>
-             <form action="/upload" enctype="multipart/form-data" method="post" id='uploadselect'>
-             	<label >Sample videos</label>
-                <select name="options" class="form-select" aria-label="Default select example" onChange={uploadName}>
-                <option selected>Open this select menu</option>
-                <option value="skate.mp4">Skate and Fall</option>
-                <option value="slip.mp4">Run and Fall</option>
-                <option value="kitchen.mp4">Fall in kitchen</option>
-                <option value="cafe_fall.mp4">Fall in workplace</option>
-                <option value="studycam.mp4">Fall experiment</option>
-              </select> 
-              {imageURL2 && <button type="submit" class="btn btn-dark" value='SubmiT' >DETECT FALL </button>}
-              </form>
-            </div>     
-        </div>
-
-        <div className='mainWrapper'>
-          <div className='mainContent'>
-            <div className='imageholder'>
-                {file_img && imageURL && <img src={imageURL} height='360' width='360' alt='Preview' crossOrigin='anonymous' ref={imageRef}/>}
-                {file_video && imageURL && <video controls src={imageURL} height='360' width='360' alt='Preview' crossOrigin='anonymous' ref={imageRef} />}
-
-            </div>                
-          </div>
-           
-        </div>
-    </div>
-  );
-}
-
-export default App;

src/index.css

@@ -1,74 +0,0 @@
-body {
-  margin: 0;
-  font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', 'Roboto', 'Oxygen',
-    'Ubuntu', 'Cantarell', 'Fira Sans', 'Droid Sans', 'Helvetica Neue',
-    sans-serif;
-  -webkit-font-smoothing: antialiased;
-  -moz-osx-font-smoothing: grayscale;
-  background-color: #8b008bcf;
-}
-
-code {
-  font-family: source-code-pro, Menlo, Monaco, Consolas, 'Courier New',
-    monospace;
-}
-
-h1 {
-  color: #FEFBEA;
-}
-
-b {
-  color: #FEFBEA;
-}
-
-.square {
-  height: 640px;
-  width: 640px;
-  background-color: #555;
-}
-
-.rcorners {
-  border-radius: 25px;
-  border: 2px solid #b494f4;
-  background-color: rgb(175, 150, 226);
-  margin: auto;
-  padding: 20px;
-  width: 80vw;
-  height: 25vh; 
-}
-.loader{
-  position: fixed;
-  top: 0;
-  left: 0;
-  width: 100vw;
-  height: 100vh;
-  display: flex;
-  justify-content: center;
-  align-items: center;
-  background-color: #96958f;
-  transition: opacity 0..75s, visibility 0.75s;
-}
-
-.loader-hidden{
-  opacity: 0;
-  visibility: hidden;
-}
-
-.loader-after{
-  content: "";
-  width: 75px;
-  height: 75px;
-  border: 15px solid #0d0d0d;
-  border-top-color: #db62f4;
-  border-radius: 50%;
-  animation: loading 0.75s ease infinite;
-}
-
-@keyframes loading {
-  from{
-    transform: rotate(0turn);
-  }
-  to{
-    transform: rotate(1turn);
-  }
-}

src/index.js

@@ -1,13 +0,0 @@
-import React from 'react';
-import ReactDOM from 'react-dom';
-import './index.css';
-import App from './App'; 
-
-ReactDOM.render(
-  <React.StrictMode>
-      <App />
-  </React.StrictMode>,
-  document.getElementById('root')
-);
-
-