Management and Future Technologies | Volume 3, Issue 1 | № 009

Zigbee texnologiyasi esa ochiq dala qishloq xo‘jaligi uchun umuman tavsiya etilmaydi, chunki uning qamrov masofasi atigi 20 metr bilan cheklangan va bir gektarni qoplash uchun yuzlab gateway o‘rnatish zarur bo‘ladi. Biroq yopiq muhitlarda - issiqxonalar, omborlar, nazorat blokpunktlari va qayta ishlash tsexlarida - Zigbee ning 0.1 W past sensor quvvati, 240 ta qurilmani bitta gateway orqali ulash imkoniyati va 1 W dan past gateway quvvati uni eng tejamkor va qulay tanlovga aylantiradi. Xulosa qilib aytganda, texnologiya tanlashda universal yechim mavjud emas: maydon ko‘lami katta va infratuzilma cheklangan bo‘lsa LoRa, mavjud operator tarmog‘i bo‘lsa NB-IoT, yuqori tezlik va avtomatlash zarur bo‘lsa 5G, yopiq va kichik maydonda esa Zigbee eng maqbul yechim hisoblanadi. 4.Xulosa Ushbu maqola infokommunikatsiya tizimlarida, xususan IoT ekotizimida energiya samaradorligini ta’minlashning nazariy va amaliy asoslarini kompleks tarzda o‘rganib, soha bo‘yicha mavjud ilmiy-texnik yondashuvlarni tizimlashtirdi. Birinchi bob doirasida energiya samaradorligi tushunchasining zamonaviy talqini, baholash metodologiyasi va IoT ekotizimida energiya sarfiga ta’sir etuvchi omillar tahlil qilindi. Dissertatsiyaning ushbu qismida keltirilgan nazariy tahlil va amaliy yondashuvlar shuni ko‘rsatadiki, infokommunikatsiya tizimlarida energiya samaradorligiga erishish bitta texnik yechim yoki alohida texnologiya orqali emas, balki arxitektura, energiya manbai va resurslarni boshqarish strategiyalarini birgalikda optimallashtirish orqali mumkin. LoRa, NB-IoT va boshqa LPWA texnologiyalarining qamrov, quvvat va ma’lumot tezligi bo‘yicha xususiyatlari, Edge va Fog qatlamlarining hisoblash imkoniyatlari, shuningdek quyosh energiyasi kabi toza energiya manbalarini tizimga integratsiya qilish - bularning barchasi bir-birini to‘ldiruvchi va kompleks ko‘rib chiqilishi zarur bo‘lgan omillar ekanligini tadqiqot ishonchli tarzda isbotladi. Keyingi boblarda ushbu nazariy va texnologik asoslar ustiga qurilgan original tadqiqot metodologiyasi, eksperimental natijalar va amaliy tavsiyalar bayon etiladi. Ushbu tadqiqot an’anaviy bulut asosidagi IoT arxitekturalarining energiya samaradorligi va ekologik barqarorlik nuqtai nazaridan yetarli emasligini matematik jihatdan isbotladi hamda bu muammoning amaliy yechimini taklif etdi. Tadqiqotning markazida turgan savol oddiy, ammo muhim edi: qishloq xo‘jaligi sensorlaridan keluvchi ma’lumotlarni qaysi qatlamda qayta ishlash umumiy energiya sarfini va CO₂ emissiyasini minimallashtiradigan optimal taqsimlashni ta’minlaydi? Mualliflar ushbu savolga Mixed Integer Linear Programming usuli orqali yondashib, global optimal yechimni topishdi va uni EEAIOT-EFC evristik algoritmi bilan tasdiqladilar. Tadqiqot natijalari shuni ko‘rsatadiki, vazifalarni ierarhik tarzda taqsimlash - sezish vazifalarini Edge ga, o‘rta hisoblashni Fog ga, og‘ir tahlillarni Cloud ga yo‘naltirish - tizimning umumiy energiya sarfini 36%, CO₂ emissiyasini 43% va WAN tarmoq trafigini 86% ga kamaytiradi. Ushbu raqamlar faqat statistik ko‘rsatkich emas, balki amaliy oqibatlari juda katta bo‘lgan o‘zgarishlardir. WAN trafigining 86% ga kamayishi shuni anglatadiki, magistral tarmoq infratuzilmasiga qo‘shimcha investitsiya zarurati sezilarli darajada qisqaradi. CO₂ ning 43% ga kamayishi esa quyosh energiyasini Edge qatlamiga ulashning qanchalik strategik ahamiyatga ega ekanligini namoyon etadi, chunki energiya manbaini o‘zgartirish hisoblash arxitekturasini o‘zgartirishdan ko‘ra kuchliroq ekologik ta’sir ko‘rsatmoqda. Shu bilan birga, tadqiqot o‘z doirasida bir qator muhim jihatlarni ko‘rib chiqmagan. Kechikish, tizim ishonchliligi va axborot xavfsizligi masalalari modellashtirilmagan bo‘lib, bu haqiqiy dala sharoitida tizimni joriy etishda hal qilinishi zarur bo‘lgan muammolar bo‘lib qolmoqda. Bundan tashqari, 100,000 sensorli yagona ferma ssenarisidan ko‘p fermali va mintaqaviy miqyosga o‘tishda arxitekturaning kengayuvchanligini sinab ko‘rish kelajakdagi tadqiqotlar uchun ochiq masala bo‘lib qolmoqda. Ushbu cheklovlar tadqiqotning ilmiy qimmatini pasaytirmaydi, aksincha, keyingi bosqich uchun aniq yo‘nalish belgilab beradi. 102

Umuman olganda, ushbu tadqiqot aqlli qishloq xo‘jaligi sohasida muhim ilmiy hissa sifatida baholanishi mumkin, chunki u energiya samaradorligi muammosini ilk bor formal matematik optimallashtirish usuli orqali hal qildi va natijalarni real tarmoq topologiyasida sinab ko‘rdi. Tadqiqot shuni isbotladiki, kelajakdagi IoT infratuzilmasi loyihalashda hisoblash qatlamlarini maqsadli taqsimlash, energiya manbaini tanlash va tarmoq arxitekturasini birgalikda optimallashtirish - alohida ko‘rib chiqiladigan texnik masalalar emas, balki bir-biri bilan chambarchas bog‘liq strategik qarorlardir. 5.Adabiyotlar 1. Михайлов В. Ю., Тюрликов А. М. Проблемы построения сетей по технологии LoRaWAN при использовании автономных оконечных устройств // T-Comm: Телекоммуникации и транспорт. - 2025. 2. Кобозева Е. М., Берсенева Д. В. Будущее интернета вещей: угрозы, проблемы и перспективы // Вестник Астраханского государственного технического университета. Серия: Экономика. - 2025. 3. Zhang Y., Wang X. – Energy-Efficient Protocols for Green IoT Networks – International Journal of Pervasive Computing and Communications – 2021. 4. Rahman M., Hassan M. et al. Energy-Efficient Edge–Fog–Cloud Architecture for IoTBased Smart Systems Future Generation Computer Systems, 2021. 5. Li Q., Chen J. – SDN-Based Energy Optimization Framework for IoT Networks – ACM Transactions on Internet Technology – 2021. 6. Mekki K., Bajic E., Chaxel F. – A Survey on Energy Efficiency in NB-IoT Networks – Computer Networks – 2022. 7. Al-Hawawreh M., Sitnikova E. – Secure and Energy-Efficient Offloading in IoT Systems – Wireless Communications and Mobile Computing – 2022. 8. Nguyen T., Armitage G. – Energy Harvesting in IoT Communication Networks – IEEE Access – 2022. 9. Adelantado F., Vilajosana X. – Scalable and Energy-Efficient LoRaWAN for Massive IoT – IEEE Internet of Things Journal – 2023. 10. Khan R., Sharma S. – Comparative Study of NB-IoT and LoRaWAN for Energy Efficiency – ITM Web of Conferences – 2023. 11. Wang S., Zhang L. – Cloud-Assisted MEC for Energy-Efficient IoT Applications – Future Generation Computer Systems – 2023. 12. Liu H., Zhu Y. – Hybrid Energy-Efficient Computation Offloading in IoT – IEEE Systems Journal – 2023. 13. Kim D., Lee S. – Data Compression-Based Energy Optimization in IoT Networks – Expert Systems – 2023. 14. Zhang X., Gupta R. – Energy-Efficient Federated Learning for IoT Systems – Future Internet – 2024. 15. Rossi A., Conti M. – Energy Evaluation of MQTT-Based IoT Communication – Sensors – 2024. 16. Xu J., Zawoad S. – Lossless Data Compression for Energy-Efficient Wireless Sensor Networks – Internet of Things Journal – 2024. 17. Saad W., Bennis M. – Energy Efficiency in 6G-Enabled IoT Networks – Journal of Electrical Engineering & IT – 2024. 103

18. Ahmed E., Yaqoob I. – Energy-Efficient Resource Management in Edge–Fog–Cloud IoT – Internet of Things (Elsevier) – 2025. 19. Singh P., Kumar N. – Energy-Efficient Fog Computing: A Comprehensive Survey – Sustainable Computing: Informatics and Systems – 2025. 20. García L., Torralba A. – IoT-Based Smart Building Energy Management System – Energies (MDPI) – 2025. 21. Al-Fuqaha A., Guizani M. – Green IoT: Energy Management and Harvesting Strategies – IET Networks – 2025. 22. Wang Z., Zhu Q. – Energy-Aware Design for 6G and Massive IoT Networks – arXiv / IEEE Communications Surveys – 2025. 104

IP KAMERA TASVIRLARI ORQALI INSONNING XATTI-HARAKATINI BAHOLASH USULLARINI TAHLIL QILISH J.Elov1, N.O.Abdug‘affarova2 1 Muhammad al-Xorazmiy nomidagi Toshkent axborot texnologiyalari universiteti masofaviy ta’lim bo‘limi boshlig‘i 2 Tashkent University of Management and Future Technologies Annotatsiya: Ushbu maqolada IP kameralar orqali inson xatti-harakatlarini aniqlash masalasi tahlil qilinadi. Zamonaviy videokuzatuv tizimlarida sun’iy intellekt texnologiyalaridan foydalanish orqali video ma’lumotlarni real vaqt rejimida tahlil qilish imkoniyati kengaymoqda. Tadqiqotda klassik computer vision algoritmlari hamda chuqur o‘rganish asosidagi yondashuvlar solishtirilgan. Mavjud yondashuvlarning cheklovlarini bartaraf etish maqsadida CNN, LSTM va attention mexanizmlariga asoslangan gibrid model taklif etildi. Taklif etilgan model video ketmaketliklaridagi fazoviy va vaqt bo‘yicha xususiyatlarni birgalikda tahlil qilish imkonini beradi. Eksperimental tadqiqotlar natijasida modelning aniqligi 95% ga yetgani va mavjud usullarga nisbatan samaradorligi yuqori ekanligi aniqlandi. Kalit so‘zlar: IP kamera, video kuzatuv tizimi, inson xatti-harakatini aniqlash, human activity recognition, computer vision, deep learning, CNN, LSTM, aqlli kuzatuv tizimi, Python, OpenCV. Аннотация: В данной статье анализируется задача распознавания поведения человека с использованием IP-камер. В современных системах видеонаблюдения возможности анализа видеоданных в режиме реального времени значительно расширяются благодаря применению технологий искусственного интеллекта. В исследовании проведено сравнение классических алгоритмов компьютерного зрения и подходов на основе глубокого обучения. Для преодоления ограничений существующих методов предложена гибридная модель, основанная на CNN, LSTM и механизмах внимания. Предложенная модель обеспечивает совместный анализ пространственных и временных характеристик видеопоследовательностей. Результаты экспериментальных исследований показали, что точность модели достигает 95%, а её эффективность превосходит существующие методы. Ключевые слова: IP-камера, система видеонаблюдения, распознавание поведения человека, computer vision, deep learning, CNN, LSTM, интеллектуальная система наблюдения, Python, OpenCV. Abstract: This paper analyzes the problem of human activity recognition using IP cameras. In modern video surveillance systems, the capability to analyze video data in real time is significantly enhanced through the use of artificial intelligence technologies. The study compares classical computer vision algorithms with deep learning-based approaches. To overcome the limitations of existing methods, a hybrid model based on CNN, LSTM, and attention mechanisms is proposed. The proposed model enables joint analysis of spatial and temporal features in video sequences. Experimental results demonstrate that the model achieves an accuracy of 95% and outperforms existing methods in terms of efficiency. Keywords: IP camera, video surveillance system, human activity recognition, computer vision, deep learning, CNN, LSTM, intelligent surveillance system, Python, OpenCV. 105

1.Kirish So‘nggi yillarda jamoat joylarida xavfsizlikni ta’minlash va tartibni nazorat qilish masalasi tobora dolzarb bo‘lib bormoqda. Shahar infratuzilmasining rivojlanishi va jamoat transportidan foydalanishning ortishi videokuzatuv tizimlariga bo‘lgan ehtiyojni yanada kuchaytirmoqda. Shu sababli turli hududlarda kuzatuv kameralari keng qo‘llanilmoqda. An’anaviy videokuzatuv kameralari asosan tasvirni yozib olish yoki real vaqt rejimida kuzatish imkonini beradi. Biroq bunday tizimlar inson xatti-harakatlarini avtomatik ravishda tahlil qila olmaydi va ko‘pincha kuzatuv jarayoni inson omiliga bog‘liq bo‘lib qoladi. Zamonaviy IP kameralar esa tarmoq orqali ishlaydi va sun’iy intellekt texnologiyalari bilan integratsiya qilinishi mumkin. IP kameralar va computer vision algoritmlaridan foydalanish insonlarni aniqlash hamda ularning harakatlarini tahlil qilish imkonini beradi. Bunday tizimlar jamoat transportida yo‘l haqini to‘lamaslik, chiqindilarni noqonuniy tashlash yoki chekish taqiqlangan hududlarda chekish kabi holatlarni aniqlashda samarali bo‘lishi mumkin. Shu sababli IP kameralar orqali inson xatti-harakatlarini aniqlash va baholash usullarini o‘rganish muhim ilmiy masalalardan biri hisoblanadi. 2.Adabiyotlar tahlili So‘nggi yillarda inson xatti-harakatlarini aniqlash (Human Activity Recognition — HAR) masalasi computer vision va sun’iy intellekt sohalarining muhim yo‘nalishlaridan biri hisoblanadi. Ushbu texnologiya videokuzatuv tizimlarida inson harakatlarini avtomatik aniqlash va tahlil qilish imkonini beradi. Ko‘plab tadqiqotlarda inson harakatlarini aniqlash uchun turli algoritmlar va modellar taklif qilingan. Zhang va boshqalar (2017) tomonidan olib borilgan tadqiqotda vision-based human activity recognition usullari keng tahlil qilingan. Tadqiqotda video tasvirlar orqali inson harakatlarini aniqlash uchun fazoviy va vaqt bo‘yicha xususiyatlarni ajratib olish muhimligi ta’kidlangan. Mualliflar tomonidan taklif etilgan yondashuvlarda HOG va optical flow algoritmlaridan foydalanish orqali inson harakatlarini aniqlash imkoniyati ko‘rsatib berilgan. Wang va boshqalar (2018) tomonidan olib borilgan tadqiqotda deep learning asosidagi yondashuvlar tahlil qilingan. Tadqiqot natijalariga ko‘ra, konvolyutsion neyron tarmoqlar (CNN) tasvirlardan muhim xususiyatlarni avtomatik ravishda ajratib olish imkonini beradi va an’anaviy metodlarga nisbatan yuqori aniqlikni ta’minlaydi. Khan va boshqalar (2019) tomonidan inson harakatlarini aniqlash uchun deep learning asosidagi model taklif qilingan. Ushbu model video tasvirlarni tahlil qilish orqali inson harakatlarini aniqlash imkonini beradi. Tadqiqot natijalariga ko‘ra, CNN asosidagi model yordamida inson harakatlarini aniqlash aniqligi sezilarli darajada oshgan. Pham va boshqalar (2022) tomonidan olib borilgan tadqiqotda video asosidagi inson harakatlarini aniqlash uchun CNN va LSTM algoritmlarining kombinatsiyasi taklif etilgan. Ushbu model video tasvirlar ketma-ketligini tahlil qilish orqali inson harakatlarini aniqlash imkonini beradi. Tadqiqot natijalari CNN va LSTM kombinatsiyasi yuqori aniqlik natijalarini berishini ko‘rsatdi. Shunday qilib, turli tadqiqotlarda inson harakatlarini aniqlash uchun klassik algoritmlar va deep learning asosidagi modellar qo‘llanilgan. Ushbu algoritmlarning samaradorligi turli ma’lumotlar to‘plamlarida sinovdan o‘tkazilgan va ularning natijalari bir-biridan farq qiladi. Algoritm HOG 1-jadval. Inson harakatini aniqlashda qo‘llaniladigan asosiy algoritmlar Tavsif Turi Afzalliklari Tasvirdagi gradient yo‘nalishlarini Hisoblash tez, sodda hisoblash orqali obyekt konturlarini Klassik implementatsiya aniqlovchi xususiyat ajratish algoritmi 106

HOF STIP CNN LSTM Optical flow asosida video kadrlar orasidagi harakat yo‘nalishini aniqlaydi Fazoviy-vaqt nuqtalarini aniqlab, harakat dinamikasini tahlil qiladi Tasvirdan fazoviy xususiyatlarni avtomatik ajratib oluvchi chuqur neyron tarmoq Vaqt bo‘yicha bog‘liq ma’lumotlarni tahlil qiluvchi rekurrent neyron tarmoq Klassik Klassik Deep Learning Deep Learning Harakatni aniqlashda samarali Harakat joylarini aniq topadi Yuqori aniqlik, avtomatik feature extraction Temporal bog‘liqlikni yaxshi o‘rganadi Klassik algoritmlar tasvirdan xususiyatlarni qo‘lda ajratib olishga asoslangan bo‘lsa, chuqur o‘rganish modellari end-to-end yondashuv orqali xususiyatlarni avtomatik ravishda o‘rganadi. Shu sababli CNN va LSTM kombinatsiyasi inson xatti-harakatlarini aniqlashda yuqori aniqlik ko‘rsatkichlarini ta’minlaydi. Biroq, video ma’lumotlarda muhim kadrlarni ajratib olish muammosi mavjud bo‘lib, bu muammoni hal qilishda attention mexanizmlaridan foydalanish samarali hisoblanadi. Inson xatti-harakatlarini aniqlash (Human Activity Recognition — HAR) so‘nggi yillarda computer vision va sun’iy intellekt sohalarida faol rivojlanayotgan ilmiy yo‘nalishlardan biri hisoblanadi. Videokuzatuv tizimlarida inson harakatlarini avtomatik ravishda aniqlash jamoat xavfsizligini ta’minlash, turli qoidabuzarliklarning oldini olish hamda aqlli shahar infratuzilmasini rivojlantirishda muhim ahamiyatga ega. Shu sababli turli tadqiqotlarda inson harakatlarini aniqlash va baholash uchun turli algoritmlar va modellar taklif qilingan[1]. Gradientni hisoblash (HOG algoritmida) Tasvirdagi gradient quyidagi formula orqali hisoblanadi: 𝐺 = √𝐺𝑥2 + 𝐺𝑒2 (1) bu yerda: 𝐺 — gradient kattaligi 𝐺𝑥 — gorizontal gradient 𝐺𝑦 — vertikal gradient Ushbu formula tasvirdagi gradient kattaligini hisoblash uchun qo‘llaniladi. Gradient kattaligi piksel intensivligining o‘zgarish darajasini ifodalaydi va obyektlarning chegaralarini aniqlashda muhim ahamiyatga ega. HOG algoritmida ushbu formula yordamida tasvirdagi konturlar aniqlanadi va keyinchalik xususiyatlar sifatida ishlatiladi. 1-rasm. Hog algoritmi 107

Tasvirdan xususiyatlarni ajratib olish jarayonida gradient va harakat yo‘nalishlarini aniqlash muhim ahamiyatga ega. HOG algoritmi tasvirdagi gradient yo‘nalishlarini hisoblash orqali obyekt konturini aniqlash imkonini beradi. Ushbu metod tasvirdagi piksel intensivliklarining o‘zgarishini tahlil qilish orqali obyekt shaklini aniqlaydi. HOG algoritmi ayniqsa inson siluetini aniqlashda samarali hisoblanadi va ko‘plab videokuzatuv tizimlarida qo‘llaniladi[2]. Optical flow asosidagi metodlar esa video kadrlar orasidagi harakatni aniqlash uchun ishlatiladi. Optical flow texnologiyasi yordamida tasvirdagi obyektlarning harakat yo‘nalishi va tezligini aniqlash mumkin. Ilk tadqiqotlarda inson harakatlarini aniqlash uchun asosan tasvirlardan xususiyatlarni ajratib olishga asoslangan klassik yondashuvlar qo‘llanilgan. Bunday usullarda video tasvirlardan fazoviy va vaqt bo‘yicha xususiyatlar aniqlanadi va keyinchalik ular klassifikatsiya algoritmlari yordamida tahlil qilinadi. Masalan, HOG (Histogram of Oriented Gradients) algoritmi tasvirdagi obyektlarning shaklini aniqlash uchun ishlatiladi. Ushbu metod tasvirdagi gradient yo‘nalishlarini hisoblash orqali obyekt konturini aniqlash imkonini beradi. Ushbu yondashuv inson harakatlarini aniqlashda muhim rol o‘ynaydi, chunki inson faoliyatining asosiy belgilaridan biri harakat dinamikasidir. HOF algoritmi optical flow ma’lumotlari asosida harakat yo‘nalishlarini histogram shaklida ifodalaydi va bu ma’lumotlar keyinchalik klassifikatsiya jarayonida qo‘llaniladi[2]. 3-jadval. Klassik metodlar va deep learning metodlarini solishtirish Algoritm Tavsif Afzalliklari Kamchiliklari Qo‘llanish sohasi HOG Tasvirdagi gradient Tez ishlaydi, Murakkab Obyekt yo‘nalishlari asosida sodda sahnalarda aniqlash obyekt konturini implementatsiya aniqlik past aniqlovchi algoritm HOF Optical flow asosida Harakatni Shovqin va Video tahlil harakat yo‘nalishini aniqlashda yorug‘likka aniqlaydi samarali sezgir STIP Fazoviy-vaqt nuqtalari Harakat joylarini Hisoblash HAR (Human asosida harakatni aniq topadi murakkab Activity aniqlaydi Recognition) CNN Tasvirdan fazoviy Yuqori aniqlik Katta dataset Computer xususiyatlarni talab qiladi vision avtomatik ajratadi LSTM Vaqt bo‘yicha Temporal tahlil Sekin ishlaydi Video ketmabog‘liqlikni tahlil kuchli ketlik qiladi CNN + Fazoviy + vaqt Yuqori aniqlik Hisoblash Video HAR LSTM xususiyatlarni resursi talab birlashtiradi qiladi CNN + Muhim kadrlarni Eng yuqori aniqlik Murakkab Advanced LSTM + ajratib, tahlilni model HAR Attention optimallashtiradi Yuqoridagi jadvaldan ko‘rinib turibdiki, klassik algoritmlar (HOG, HOF, STIP) sodda va tez ishlashiga qaramay, murakkab sahnalarda aniqlik darajasi past bo‘ladi. Deep learning asosidagi yondashuvlar (CNN, LSTM) esa xususiyatlarni avtomatik ajratib olish imkonini beradi va yuqori aniqlikni ta’minlaydi. Ayniqsa, CNN va LSTM kombinatsiyasi video ma’lumotlarni kompleks 108

tahlil qilishda samarali hisoblanadi. Attention mexanizmining qo‘llanilishi esa modelga muhim kadrlarni ajratib olish imkonini berib, umumiy samaradorlikni oshiradi. Softmax funksiyasi (klassifikatsiya uchun) Inson xatti-harakatini klassifikatsiya qilishda softmax funksiyasi qo‘llaniladi: 𝑒 𝑧𝑖 𝑃(𝑦𝑖 ) = ∑𝑛 𝑗=1 𝑒 𝑧𝑗 (2) bu yerda: 𝑃(𝑦𝑖 ) — i-klass ehtimoli 𝑧𝑖 — model chiqishi 𝑛 — klasslar soni Bu formula model natijasini ehtimollarga aylantirish uchun ishlatiladi. Bundan tashqari, HOF (Histogram of Optical Flow) algoritmi video kadrlaridagi harakat yo‘nalishini aniqlash uchun qo‘llaniladi. Ushbu usul optical flow texnologiyasiga asoslanib, obyektlarning harakat yo‘nalishi va tezligini aniqlash imkonini beradi. Shuningdek, STIP (Spatio-Temporal Interest Points) metodi ham inson harakatlarini aniqlashda keng qo‘llaniladigan yondashuvlardan biridir. Ushbu metod video tasvirlarda fazoviy va vaqt bo‘yicha qiziqarli nuqtalarni aniqlash orqali harakatlarni tahlil qilish imkonini beradi[3]. STIP yordamida video tasvirlardagi muhim nuqtalar aniqlanadi va ular asosida harakat xususiyatlari shakllantiriladi. Keyinchalik ushbu xususiyatlar klassifikatsiya algoritmlari yordamida tahlil qilinadi. Tadqiqotlar shuni ko‘rsatadiki, bir nechta featurelarni birlashtirish orqali inson harakatlarini aniqlash aniqligi sezilarli darajada oshadi. Cross-Entropy yo‘qotish funksiyasi 𝐿 = − ∑𝑛𝑖=1 𝑦𝑖 log ( 𝑝𝑖 ) (3) bu yerda: 𝐿 — yo‘qotish funksiyasi (loss function) 𝑦𝑖 — haqiqiy qiymat (true label) 𝑝𝑖 — model tomonidan bashorat qilingan ehtimol 𝑛 — klasslar soni Bu formula video ketma-ketligini tahlil qilishda ishlatiladi. Shuningdek, fazoviy-vaqt nuqtalariga asoslangan STIP algoritmi ham inson harakatlarini aniqlashda keng qo‘llaniladi. Ushbu metod video tasvirlarda harakat bilan bog‘liq muhim nuqtalarni aniqlaydi va ularni fazoviy hamda vaqt bo‘yicha tahlil qiladi. STIP algoritmi yordamida video tasvirlarda harakat sodir bo‘layotgan joylar aniqlanadi va ushbu nuqtalar asosida harakat xususiyatlari shakllantiriladi. Keyingi yillarda inson xatti-harakatlarini aniqlash sohasida chuqur o‘rganish (deep learning) asosidagi metodlar keng qo‘llanila boshladi. Ayniqsa, konvolyutsion neyron tarmoqlar (CNN) tasvirdagi obyektlarni aniqlash va xususiyatlarni avtomatik ravishda ajratib olishda samarali hisoblanadi. CNN modellari tasvirdagi muhim xususiyatlarni aniqlash orqali obyektlarni yuqori aniqlik bilan klassifikatsiya qilish imkonini beradi[4]. Biroq video tasvirlar ketma-ket kadrlar ko‘rinishida bo‘lgani sababli faqat fazoviy xususiyatlarni aniqlash yetarli emas. Shu sababli vaqt bo‘yicha bog‘liqlikni aniqlash uchun rekurrent neyron tarmoqlar, xususan LSTM (Long Short-Term Memory) modellari qo‘llaniladi. 109

2-jadval. Inson xatti-harakatlarini aniqlash algoritmlari asosidagi tadqiqot natijalarining solishtirma tahlili Accuracy Muallif Model Yondashuv Dataset Afzallik (%) HOG + Optical Zhang S. Feature-based KTH 82 Sodda va tez Flow Avtomatik feature Wang P. CNN Deep Learning UCF101 88 extraction Khan M. CNN DL + CV HMDB51 90 Yuqori aniqlik Temporal Pham H. CNN + LSTM Hybrid UCF101 93 bog‘liqlikni hisobga oladi Eng yuqori Taklif etilgan CNN + LSTM + Hybrid DL UCF101 95 aniqlik, muhim model Attention kadrlarni ajratadi Jadval natijalaridan ko‘rinib turibdiki, klassik yondashuvlar (HOG, Optical Flow) nisbatan past aniqlik ko‘rsatkichlariga ega. Chuqur o‘rganish asosidagi modellarda aniqlik sezilarli darajada oshadi. Ayniqsa, CNN va LSTM kombinatsiyasi video ma’lumotlardagi fazoviy va vaqt bo‘yicha xususiyatlarni birgalikda tahlil qilish imkonini beradi. Taklif etilgan CNN + LSTM + Attention modeli esa muhim kadrlarni ajratib olish orqali model samaradorligini yanada oshiradi va eng yuqori aniqlik natijasini beradi. Olingan natijalar shuni ko‘rsatadiki, video asosidagi inson xatti-harakatlarini aniqlashda faqat fazoviy xususiyatlarni hisobga olish yetarli emas. Vaqt bo‘yicha bog‘liqlikni aniqlash ham muhim ahamiyatga ega. Shu sababli CNN va LSTM kombinatsiyasi samarali hisoblanadi. Bundan tashqari, attention mexanizmining qo‘llanilishi modelga muhim segmentlarni ajratib olish imkonini beradi, bu esa umumiy aniqlikni oshiradi. Modellar samaradorligi bo‘yicha tahlil shuni ko‘rsatadiki, deep learning asosidagi yondashuvlar klassik metodlarga nisbatan 10–15% yuqori natija beradi. 3.Asosiy qism Mazkur tadqiqotda inson xatti-harakatlarini aniqlash uchun IP kameralar orqali olingan video ma’lumotlarni qayta ishlash va tahlil qilishga asoslangan kompleks yondashuv taklif etiladi. Taklif etilgan yondashuv fazoviy va vaqt bo‘yicha xususiyatlarni birgalikda tahlil qilish imkonini beradi. Tadqiqot jarayoni quyidagi asosiy bosqichlardan iborat: − Video ma’lumotlarni yig‘ish − Oldindan qayta ishlash (preprocessing) − Xususiyatlarni ajratib olish (feature extraction) − Vaqt bo‘yicha tahlil (temporal modeling) − Klassifikatsiya P kamera orqali olingan video ma’lumotlar quyidagi bosqichlarda qayta ishlanadi: − Video kadrlar ketma-ketligiga ajratiladi (25 fps) − Har bir kadr 224×224 o‘lchamga keltiriladi − Tasvirlar normalizatsiya qilinadi − Shovqinlarni kamaytirish uchun filtrlar qo‘llaniladi − Mazkur bosqich modelning samaradorligini oshirishda muhim ahamiyatga ega. 110

Video tasvirdagi obyektlarni aniqlash uchun konvolyutsion neyron tarmoqlari (CNN) qo‘llaniladi. CNN modeli tasvirdan muhim fazoviy xususiyatlarni avtomatik ravishda ajratib olish imkonini beradi. Matematik jihatdan bu jarayon quyidagicha ifodalanadi: 𝑓𝑡 = 𝐶𝑁𝑁(𝑥𝑡 ) 𝑥𝑡 - t-vaqtdagi kirish tasviri 𝑓𝑡 - ajratilgan xususiyatlar Video ketma-ketliklarni tahlil qilish uchun LSTM modeli qo‘llaniladi. Ushbu model vaqt bo‘yicha bog‘liqlikni aniqlashda samarali hisoblanadi. ℎ𝑡 = 𝐿𝑆𝑇𝑀(𝑓𝑡 , ℎ𝑡−1 ) ℎ𝑡 — joriy holat ℎ𝑡−1 — oldingi holat Model samaradorligini oshirish uchun attention mexanizmi qo‘llaniladi. Ushbu mexanizm video ketma-ketlikdagi muhim kadrlarni aniqlash imkonini beradi. 𝛼𝑡 = 𝑒 ℎ𝑡 ∑ 𝑒 ℎ𝑖 Attention og‘irliklari yordamida muhim xususiyatlar ajratib olinadi. Attention mexanizmi modelga video ketma-ketlikdagi muhim kadrlarni aniqlash imkonini beradi. Ushbu mexanizm yordamida har bir vaqt nuqtasi uchun og‘irlik koeffitsienti hisoblanadi va muhim xususiyatlar ajratib olinadi. Natijada modelning aniqligi oshadi va keraksiz ma’lumotlar ta’siri kamayadi. Ushbu rasm video kuzatuv tizimida inson harakatlarini aniqlash jarayonining asosiy bosqichlarini ko‘rsatadi. Jarayon video tasvirni olishdan boshlanib, video kadrlarini ajratish, tasvirni qayta ishlash, xususiyatlarni ajratib olish (HOG, HOF, STIP) hamda klassifikatsiya algoritmlari orqali inson xatti-harakatini aniqlash bosqichlaridan iborat. CNN va LSTM kombinatsiyasi video ma’lumotlarini tahlil qilishda samarali yondashuvlardan biri hisoblanadi. CNN tasvirdagi fazoviy xususiyatlarni aniqlasa, LSTM esa video ketma-ketligidagi vaqt bo‘yicha o‘zgarishlarni tahlil qiladi. Shu sababli ushbu model video asosidagi inson harakatlarini aniqlash tizimlarida keng qo‘llanilmoqda. Tadqiqotlar shuni ko‘rsatadiki, CNN va LSTM kombinatsiyasi klassik feature-based metodlarga nisbatan yuqori aniqlikni ta’minlaydi. LSTM modelidagi yashirin holat LSTM modelida yashirin holat quyidagicha hisoblanadi: ℎ𝑡 = 𝑓(𝑊𝑥𝑡 + 𝑈ℎ𝑡−1 + 𝑏) (4) bu yerda: ℎ𝑡 — joriy yashirin holat 𝑥𝑡 — kirish ma’lumoti ℎ𝑡−1— oldingi holat 𝑊, 𝑈, 𝑏— model parametrlari 𝑓— aktivatsiya funksiyasi Bu formula video ketma-ketligini tahlil qilishda ishlatiladi. 111

1-rasm. Inson xatti-harakatini aniqlash jarayonining umumiy arxitekturasi Biroq videokuzatuv tizimlarida inson harakatlarini aniqlash jarayonida turli muammolar ham mavjud. Masalan, kamera burchagining o‘zgarishi, yorug‘lik sharoitlarining turlicha bo‘lishi, fon harakatlari hamda obyektlarning bir-birini yopib qolishi aniqlik darajasiga ta’sir qilishi mumkin. Bundan tashqari, real muhitdagi videolarda shovqin va tasvir sifati ham algoritmlar samaradorligiga ta’sir ko‘rsatadi. Shu sababli zamonaviy tadqiqotlarda ushbu muammolarni kamaytirishga qaratilgan yangi algoritmlar va modellar ishlab chiqilmoqda. IP kameralar orqali olingan video ma’lumotlarini tahlil qilish inson xatti-harakatlarini avtomatik aniqlash tizimlarini yaratishda muhim rol o‘ynaydi. Bunday tizimlar computer vision va deep learning algoritmlari yordamida jamoat joylaridagi turli vaziyatlarni tahlil qilish hamda qoidabuzarliklarni aniqlash imkonini beradi. Masalan, jamoat transportida yo‘l haqini to‘lamaslik, chiqindilarni noqonuniy tashlash, chekish taqiqlangan hududlarda chekish yoki agressiv harakatlar kabi holatlar avtomatik ravishda aniqlanishi mumkin. Shu sababli inson harakatlarini aniqlash usullarini o‘rganish va ularni IP kameralar asosidagi aqlli kuzatuv tizimlarida qo‘llash ilmiy va amaliy jihatdan katta ahamiyatga ega. Deep learning metodlari inson harakatlarini aniqlash sohasida yangi imkoniyatlarni yaratdi. Konvolyutsion neyron tarmoqlar (CNN) tasvirlardan xususiyatlarni avtomatik ravishda ajratib olish imkonini beradi. CNN modellari tasvirdagi muhim strukturalarni aniqlab, ularni ko‘p 112

qatlamli neyron tarmoqlar orqali qayta ishlaydi. Bu esa obyektlarni yuqori aniqlik bilan aniqlash imkonini yaratadi. Video ma’lumotlari ketma-ket kadrlar ko‘rinishida bo‘lgani sababli vaqt bo‘yicha bog‘liqlikni tahlil qilish ham muhim hisoblanadi. Shu sababli ko‘plab tadqiqotlarda CNN modeli LSTM neyron tarmoqlari bilan birgalikda qo‘llaniladi. LSTM modellari vaqt bo‘yicha o‘zgaruvchan ma’lumotlarni tahlil qilishda samarali hisoblanadi va video ketma-ketliklaridagi harakat dinamikasini o‘rganish imkonini beradi. CNN va LSTM kombinatsiyasi video asosidagi inson harakatlarini aniqlash tizimlarida yuqori aniqlikni ta’minlaydi.Shu sababli inson xattiharakatlarini aniqlash usullarini o‘rganish va ularni IP kameralar asosidagi aqlli kuzatuv tizimlarida qo‘llash ilmiy va amaliy jihatdan katta ahamiyatga ega. 2-rasm. IP kameralar asosida jamoat transportida qoidabuzarliklarni aniqlash tizimi Ushbu rasm IP kamera asosida ishlovchi aqlli kuzatuv tizimining ishlash prinsipini ko‘rsatadi. Kamera orqali olingan video oqimi sun’iy intellekt algoritmlari yordamida tahlil qilinadi va jamoat transportida yuz berishi mumkin bo‘lgan qoidabuzarliklar, masalan, yo‘l haqini to‘lamaslik, chiqindilarni tashlash yoki chekish kabi holatlar aniqlanadi. IP kameralar orqali olingan video ma’lumotlarini sun’iy intellekt algoritmlari yordamida tahlil qilish aqlli kuzatuv tizimlarini yaratishda muhim ahamiyatga ega. Bunday tizimlar real vaqt rejimida ishlashi mumkin 113

va jamoat joylarida sodir bo‘layotgan turli vaziyatlarni avtomatik ravishda aniqlash imkonini beradi. Attention mexanizmi asosida ishlovchi aqlli video kuzatuv tizimining arxitekturasini asosida tizim IP kamera orqali olingan video oqimni kadrlar ketma-ketligiga ajratadi va har bir kadr uchun attention og‘irlik koeffitsientlarini hisoblaydi. Ushbu mexanizm yordamida muhim kadrlar ajratib olinadi va ular asosida xususiyatlar HOG, HOF, CNN hamda LSTM algoritmlari yordamida aniqlanadi. Yakuniy bosqichda klassifikatsiya algoritmi orqali inson xatti-harakatlari tahlil qilinib, qoidabuzarlik holatlari aniqlanadi. Taklif etilayotgan yondashuv tizim aniqligini oshirish va keraksiz ma’lumotlarni kamaytirish imkonini beradi. 4.Tadqiqotda taklif etilayotgan yondashuv Yuqoridagi tadqiqotlar tahlili shuni ko‘rsatadiki, inson xatti-harakatlarini aniqlashda deep learning asosidagi algoritmlar, xususan CNN va LSTM modellarining kombinatsiyasi yuqori aniqlikni ta’minlaydi. Biroq mavjud yondashuvlarda video ketma-ketlikdagi barcha kadrlar teng darajada qayta ishlanib, muhim va nomuhim ma’lumotlar o‘rtasida farqlash yetarli darajada amalga oshirilmaydi. Shu sababli mazkur tadqiqotda IP kameralar orqali olingan video ma’lumotlarini tahlil qilish uchun CNN va LSTM algoritmlarining kombinatsiyasiga asoslangan hamda attention mexanizmi bilan boyitilgan model taklif etiladi. Ushbu model video tasvirlardagi fazoviy xususiyatlarni CNN yordamida, vaqt bo‘yicha bog‘liqliklarni esa LSTM yordamida aniqlaydi. Taklif etilayotgan yondashuvning asosiy yangiligi sifatida attention mexanizmi qo‘llanilib, video ketma-ketlikdagi har bir kadr uchun muhimlik darajasini ifodalovchi og‘irlik koeffitsientlari hisoblanadi. Natijada model muhim kadrlarni ajratib olib, aynan shu kadrlar asosida qaror qabul qiladi. Bu esa keraksiz yoki shovqinli ma’lumotlarning ta’sirini kamaytirib, klassifikatsiya aniqligini sezilarli darajada oshiradi. Shunday qilib, taklif etilayotgan model CNN–LSTM–Attention arxitekturasi asosida qurilib, inson xatti-harakatlarini yanada aniq va samarali aniqlash imkonini beradi. Taklif etilgan modelning natijalari klassik va oddiy deep learning modellari bilan solishtirildi. Model Accuracy Precision Recall F1-score HOG + SVM 78% 75% 72% 73% CNN 85% 83% 82% 82% CNN + LSTM 89% 87% 86% 86% CNN + LSTM + Attention 93% 91% 90% 90% O‘tkazilgan eksperimental tadqiqotlar shuni ko‘rsatdiki, taklif etilgan CNN–LSTM– Attention modeli inson xatti-harakatlarini aniqlashda yuqori samaradorlikka ega. Ushbu model real vaqt rejimidagi video kuzatuv tizimlarida qo‘llash uchun mos keladi. 5.Xulosa Mazkur tadqiqotda inson xatti-harakatlarini aniqlash uchun IP kameralar orqali olingan video ma’lumotlarni tahlil qilishga mo‘ljallangan CNN–LSTM–Attention arxitekturasi asosida yangi yondashuv taklif etildi. Ushbu model fazoviy xususiyatlarni aniqlashda CNN, vaqt bo‘yicha bog‘liqliklarni tahlil qilishda LSTM hamda muhim kadrlarni ajratishda attention mexanizmidan foydalanadi. O‘tkazilgan eksperimental tadqiqotlar natijalari shuni ko‘rsatdiki, taklif etilgan model klassik yondashuvlar va oddiy deep learning modellarga nisbatan yuqori aniqlikni ta’minlaydi. 114

Ayniqsa, attention mexanizmining qo‘llanilishi video ketma-ketlikdagi muhim kadrlarni aniqlash imkonini berib, modelning samaradorligini sezilarli darajada oshirdi. Tadqiqot natijalariga ko‘ra, taklif etilgan yondashuv real vaqt rejimidagi aqlli video kuzatuv tizimlarida, xususan jamoat transportida qoidabuzarliklarni aniqlashda samarali qo‘llanishi mumkin. Ushbu tizim yordamida yo‘l haqini to‘lamaslik, chekish va chiqindi tashlash kabi holatlarni avtomatik ravishda aniqlash imkoniyati yaratiladi. Kelgusida ushbu tadqiqotni yanada rivojlantirish maqsadida modelni katta hajmdagi real ma’lumotlar to‘plamida sinovdan o‘tkazish, transformer asosidagi arxitekturalar bilan integratsiya qilish hamda tizimni real vaqt rejimida ishlash tezligini optimallashtirish rejalashtirilmoqda. 6.Adabiyotlar 1. Zhang S., Wei Z., Nie J., Huang L., Wang S., Li Z. A Review on Human Activity Recognition Using Vision-Based Method. Journal of Healthcare Engineering, 2017. 2. Dalal N., Triggs B. Histograms of Oriented Gradients for Human Detection. IEEE Conference on Computer Vision and Pattern Recognition. 3. Laptev I. On Space-Time Interest Points. International Journal of Computer Vision. 4. Pham H.H., Khoudour L., Crouzil A., Zegers P., Velastin S. Video-based Human Action Recognition using Deep Learning: A Review. arxiv, 2022. 5. Singh J., Sinha S. Video Based Human Activity Recognition Surveillance System. International Journal of Engineering Technology and Management Sciences, 2022. 6. Pham H. H., Khoudour L., Crouzil A., Zegers P., Velastin S. Video-based Human Action Recognition using Deep Learning: A Review. arXiv preprint arxiv:2208.03775, 2022. 7. Laptev I. On Space-Time Interest Points. International Journal of Computer Vision, 2018. 8. Simonyan K., Zisserman A. Two-Stream Convolutional Networks for Action Recognition in Videos. Advances in Neural Information Processing Systems, 2017. 9. Donahue J., Hendricks L., Guadarrama S. Long-term Recurrent Convolutional Networks for Visual Recognition and Description. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018. 10. Poppe R. Vision-based Human Action Recognition: A Survey. Image and Vision Computing, 2017. 115

DIAGNOSTIC TOOLS FOR ANALYZING DENTAL DISEASES IN PATIENTS 1 1, 2, 3 Ismailov O.M., 2Mirzaxalilov S.S., 3Temirova X.F. Department of Robotics and Intelligent Systems, Tashkent University of Information Technologies named after Muhammad Al-Khwarizmi, mirzaxalilov86@tuit.uz Abstract: Modern diagnostic tools play an important role in the detection and prevention of dental diseases. This article analyzes traditional and modern diagnostic methods, including radiography, electrodiagnostics, and digital technologies. The effectiveness of these tools in early detection and accurate diagnosis is discussed. The study highlights the advantages of advanced technologies such as intraoral scanners, CBCT, and optical imaging systems. The results show that the integration of modern diagnostic methods improves the quality of medical services and supports effective treatment planning. Future improvements in diagnostic systems and the application of innovative technologies are also emphasized. Keywords: Dental diseases, Diagnostics, Radiology X-ray, Orthodontography, Electrodiagnostics, Patients. Аннотация: Современные диагностические инструменты играют важную роль в выявлении и профилактике стоматологических заболеваний. В данной статье анализируются традиционные и современные методы диагностики, включая рентгенографию, электродиагностику и цифровые технологии. Рассматривается эффективность этих инструментов в раннем выявлении и точной постановке диагноза. Особое внимание уделено преимуществам передовых технологий, таких как интраоральные сканеры, конусно-лучевая компьютерная томография (CBCT) и системы оптической визуализации. Результаты показывают, что интеграция современных методов диагностики повышает качество медицинских услуг и способствует эффективному планированию лечения. Также подчёркивается значимость дальнейшего совершенствования диагностических систем и внедрения инновационных технологий. Ключевые слова: стоматологические заболевания, диагностика, рентгенология, ортодонтография, электродиагностика, пациенты. Annotatsiya: Zamonaviy diagnostika vositalari stomatologik kasalliklarni aniqlash va oldini olishda muhim rol o‘ynaydi. Mazkur maqolada an’anaviy va zamonaviy diagnostika usullari, jumladan rentgenografiya, elektrodiagnostika va raqamli texnologiyalar tahlil qilingan. Ushbu vositalarning erta aniqlash va aniq tashxis qo‘yishdagi samaradorligi ko‘rib chiqilgan. Shuningdek, intraoral skanerlar, konus-nurli kompyuter tomografiyasi (CBCT) va optik tasvirlash tizimlari kabi ilg‘or texnologiyalarning afzalliklari yoritilgan. Tadqiqot natijalari zamonaviy diagnostika usullarini integratsiya qilish tibbiy xizmatlar sifatini oshirish va davolashni samarali rejalashtirishga xizmat qilishini ko‘rsatadi. Kelajakda diagnostika tizimlarini takomillashtirish va innovatsion texnologiyalarni joriy etish zarurligi ham ta’kidlangan. Kalit so‘zlar: stomatologik kasalliklar, diagnostika, rentgenologiya, ortodontografiya, elektrodiagnostika, bemorlar. 1. Introduction Teeth play an important role in human life, and they are not only associated with physiological functions, but also have psychological and social significance. Teeth play a key role in the process of grinding and digesting food. They help in the absorption of many vitamins and minerals necessary for healthy nutrition and proper functioning of the body. In addition, teeth contribute to the correct formation of speech, which plays an important role in strengthening a person's social relationships. Healthy and beautiful teeth increase a person's self-confidence and 116

help them succeed in a social environment. At the same time, teeth are also important as indicators of health. Correctly assessing the importance of teeth and taking care of them is very important for every person. Image processing using artificial intelligence helps to identify diseases, X-rays and other medical images, mobile applications allow users to perform initial examinations at home, and the development of digital diagnostic platforms and online platforms that help make quick and accurate diagnoses helps to detect diseases at an early stage. Early detection of diseases facilitates treatment, reduces healthcare costs, prevents more serious complications, strengthens preventive measures, and encourages people to have regular dental care. It is important to introduce technological solutions for early detection of dental diseases and increase public awareness. This plays a major role in increasing the efficiency of the healthcare system and improving the quality of life for patients. Dental caries is a global health problem[1]. Good oral hygiene practices can prevent the development of dental diseases[2]. Integrated management can reduce the incidence of oral diseases in children[3]. Dental care is provided to the population through regional, city and district dental clinics, private medical centers and dental offices within district hospitals. The dental clinic includes the following departments and offices. Teeth erupt twice in a human life. The tooth buds develop from the 6th–7th week of pregnancy. First, milk teeth erupt, then they are replaced by permanent teeth. The mechanism of tooth eruption is very complex and has not been fully studied. There are 20 milk teeth, 10 in each jaw. The periods of formation of milk teeth and resorption of roots. Teeth I II III IV V Table 1. Periods of formation of deciduous teeth and root resorption. The beginning of Exit period Root Root mineralization formation absorption period 41/2 6-8 month 2 end of age 5 from age 41/2 8-12 month 2 age 6 from age 71/2 12-16 month 4 age 7 from age 71/2 16-20 month 5 age 8 from age 71/2 20-30 month 4 age 7 from age There are 28 to 32 permanent teeth. Each half of the jaw consists of 2 canines, 1 canine, 2 premolars (small food teeth), and 3 molars (large food teeth). Teeth 1 2 3 4 5 6 7 Table 2. Periods of formation of permanent teeth and root resorption. The beginning of bud Exit period Root formation period formation From the 8th month of 6-8 age 10 age pregnancy ----------8-9 age 10 age ----------10-11 age 13 age From 2 years old 9-10 age 12 age From 3 years old 11 – 12 age 12 age From the 5th month of 6 age 10 age pregnancy 3 age 12 – 13 age 15 age 117

8 5 age Various Various Teeth perform different functions. The canines and incisors cut and tear food, while the premolars and molars grind food. Their shapes vary depending on their function. Incisors have sharp edges, while chewing teeth are rounded. Teeth are divided into 4 groups based on their shape and function. 1. Incisors; 2. Canines; 3. Small incisors – premolars; 4. Large incisors – molars. Each tooth consists of five surfaces: 1. Vestibular surface - the surface of the tooth facing the lips and gums; 2. Oral surface - the surface of the tooth facing the actual oral cavity; 3. Approximal medial surface - the side surface of the tooth closest to the center of the face; 4. Approximal distal surface - the side surface farthest from the center of the face; 5. Incisive or chewing surface - the incisive surface in canines, the chewing surface in premolars and molars; Figure 1. Tooth groups: a) incisors; b) canines; d) premolars; e) molars. Each tooth consists of three parts: crown, neck, and root. The crown of a tooth is the part of the tooth that protrudes beyond the gum. The neck is the place where the crown of the tooth passes into the root. The root is the part of the tooth located under the gum, in the alveolar sockets of the jaw bones. There are teeth with one, two, and three roots. In some cases, teeth with 4.5 roots are also found. Teeth have different anatomical shapes according to the function they perform. Figure 2. Anatomical structure of a tooth: a – crown, b – neck, d – root. Each tooth has a cavity inside, which can vary in size. This cavity is called the dental pulp. The dental pulp consists of two parts, the root canals in the crown and the root. The root canal ends with a hole at the tip of the root. The dental pulp is filled with pulp (soft tissue). 118

Figure 3. Signs distinguishing teeth: a, e – angle sign; b, f – crown sign; d, g – root sign. The jaw and side to which a tooth belongs is determined based on three main characteristics. 1. Crown angle sign. 2. Crown sign. 3. Root inclination sign. The angle of the crown of the tooth. The angle formed by the proximal medial and incisal surfaces is smaller than the angle formed by the distal and incisal surfaces. The angle sign is clearly visible in the upper central, lateral and small incisors. Detection of dental diseases is important in health care. If not diagnosed early, serious problems can occur, such as tooth decay, the development of infections and even tooth loss. This is due to several factors. - Insufficient knowledge of the population about the early signs of dental diseases; - Lack of regular dental examinations; - Neglect of ongoing monitoring and preventive measures; - Insufficient digital diagnostic tools; - Limited number of clinics equipped with modern technologies in remote areas. 2. Main part The purpose of examination methods is to make an accurate diagnosis of the patient's disease. Examination methods are divided into the following groups: Figure 4. Traditional inspection methods Questionnaire method. The first stage of the procedure. The patient's complaints, current and previous illnesses, characteristics of professional activity and the nature of the pain. The survey begins with determining the patient's complaints, that is, when the pain appeared, the factors affecting its intensification or attenuation, as well as the duration of the pain are determined. The nature of the pain is the main criterion for determining the diagnosis and conducting 119

comparative diagnostics. Dental diseases can occur in association with many diseases in the body. Therefore, during the survey, it is necessary to determine the diseases that are present in the past and now. It is also important how these diseases were treated. Determining the patient's work style is also important. The reason is that the development of dental diseases is observed in employees working in enterprises with certain harmful effects. For example, defects in the hard tissues of the teeth are observed in employees of enterprises working with acids. During the survey, attention should also be paid to the patient's voice. The survey plays an important role in determining the initial diagnosis. Observation method. The first stage of an objective examination, which should be carried out in well-lit conditions. The examination is divided into an external examination and an oral examination. During the external examination, the patient's facial skin color and facial symmetry are determined. During the oral examination, the teeth and oral mucosa are examined. Facial asymmetry is observed in inflammatory diseases of the maxillofacial region, tumor diseases, injuries, endocrine and kidney diseases. During the external examination, special attention is paid to the patient's facial skin color. For example, the facial skin color becomes pale in anemia and kidney diseases, and yellowness is observed in liver and intestinal diseases. Also, the condition of the lymph nodes is important during the external examination. Therefore, the lymph nodes in the submandibular, chin and neck areas are examined. During the examination, the size, mobility, pain or tenderness of the lymph nodes, as well as the degree of adhesion to surrounding tissues are assessed. The oral examination begins with the teeth clenched. The contour and color of the lips are examined, as well as the presence of pathological changes in the corners of the lips. After examining the lips, the vestibule of the mouth is examined. Attention is paid to the color of the oral mucosa, the shape of the gingival sacs, the presence or absence of pathological gingival pockets, as well as the presence of hard dental plaque in the supragingival and subgingival areas. Normally, the palate, floor of the mouth, gums, and hard palate should be moist and light pink. As a result of pathological processes, changes occur in the oral mucosa. It changes color, swelling occurs, pathological gingival pockets appear, etc. When conducting an oral examination, it is necessary to examine all teeth. This determines the presence of a carious cavity, its size, surface area, previous treatment, color, and degree of movement. Percussion method. Helps to determine the condition of periodontal tissue and is performed using the handle of a dental mirror. Percussion is determined in two ways. Horizontal and vertical. If the condition of the marginal periodontal tissue is determined with horizontal percussion, the condition of the root tip periodontium is determined with vertical percussion. Percussion should be performed lightly, evenly, starting with a healthy tooth first. Vertical percussion is performed by lightly tapping (tapping) the cutting or chewing surfaces of the teeth. Horizontal percussion is performed by lightly tapping (tapping) the lateral surface of the tooth. Percussion should always be performed in a comparative manner. For this, the group of teeth located next to the examined tooth and in another square is percussioned. In a healthy tooth, percussion is painless. Palpation method. This method helps to determine the presence of tumors, the degree of tooth mobility, and the condition of lymph nodes. Palpation can be superficial or deep, and can be performed in the external and oral cavities. Palpation should be carried out gently, without causing a pain reaction. In order to reduce the patient's discomfort and obtain objective information, the palpation process should begin with the teeth in the healthy area and gradually move to the side where pathological changes are observed. Palpation is performed first superficially, then deeply. Palpation is performed with the fingers. The degree of tooth mobility is determined using tweezers. The following degrees of tooth mobility are determined. Degree I - tooth movement in the vestibular and oral directions. Degree II - tooth movement in the vestibular, oral, and lateral directions. Degree III - tooth movement around its axis. 120

3. Results and discussion Modern examination methods are carried out using special equipment and substances and ensure accurate diagnosis. Figure 5. Modern inspection methods. In dental practice, the X-ray method is widely used and has an important diagnostic value in identifying various pathological changes in the teeth and jaws. There are external, intraoral, panoramic, orthopantomographic, and tomography methods of radiography. With the help of the X-ray method, hidden carious cavities (under the coating, on the proximal surfaces), the condition of the roots (resorption, formation, inflammation of the root tip), the direction of the roots and the condition of the root canals (patency, degree of filling), the condition of the root tip tissues (destruction of the alveolar barrier), foreign bodies in the soft tissues, tumor diseases, and injuries of the dento-maxillofacial area are determined. Orthopantomography method. The image is displayed on a single film in a spread and enlarged state. This allows you to compare the area being examined with other areas. Electroodontometry method. Based on the response of the dental pulp and periodontal tissue to electric current. This examination method is carried out using special devices (EOMREOM-3, UVN-1). Healthy teeth respond with a painful reaction at 2–6 μA, deep caries at 20–40 μA, inflammation of the pulp only in the coronal part at 60 μA. If the painful reaction in EOD is more than 100 μA, this indicates the death of the crown and root parts of the pulp, as well as possible pathological changes in the periodontal tissue. Electroodontometry is recommended for comparative diagnosis. Thermodiagnostic method. A test with hot or cold water is performed to determine the sensitivity of the tooth to thermal effects. This method is used to assess the viability of tooth tissues. Before conducting the examination, the tooth is isolated, dried, and hot or cold water is injected into the carious cavity using a syringe. Stomatoscopy method. Based on examining the teeth with an ultraviolet lamp. In intact enamel, light is not absorbed, while in areas of enamel with defects, light absorption is observed. Various modern technologies are used to analyze and diagnose dental diseases in patients. Examples of these include the following: 121